Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
SYSTEM FOR DETECTING ICE ON METAL SURFACE
Document Type and Number:
WIPO Patent Application WO/1994/009463
Kind Code:
A1
Abstract:
For detecting the presence of a substance such as ice or snow on a surface such as the wing of an aircraft, light is emitted from an unpolarized source (13). The unpolarized light passes through a linear polarizer (11) which has a vertical polarization axis. The vertically polarized light passes through a quarter wave retarder plate (12), which changes the linearly polarized light into circularly polarized light. The quarter wave retarder plate (12) has its slow and fast axes both at 45 degrees relative to the vertical axis of the linear polarizer (11). A specular surface (14) reflects the incident circularly polarized light back along path (23). If surface (14) is ice-free, the reflected light will result in an image viewed by the eye (26) alternating between dark and bright, respectively. Any ice or snow covering a portion of surface (14) will cause that portion of the image to maintain the same brightness.

Inventors:
Stern, Howard
Application Number:
PCT/US1993/010035
Publication Date:
April 28, 1994
Filing Date:
October 20, 1993
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
ROBOTIC VISION SYSTEMS, INC
Stern, Howard
International Classes:
B64D15/20; G08B19/02; (IPC1-7): G08B19/02
Foreign References:
US4701052A
US3994586A
US3801182A
US5243185A
Other References:
See also references of EP 0665972A1
Download PDF:
Claims:
WHAT IS CLAIMED:
1. l. Apparatus for detecting the presence of a dielectric substance such as ice or snow on a surface which specularly reflects light comprising: means for transmitting light over a path to said surface, means for receiving the light reflected from said surface, and optical means in the light path between one of the transmitting means and the surface and the surface and the receiving means alternating between an optical isolator and an optical nonisolator state, the reflected light from the specularly reflecting portion of the surface correspondingly alternating from one brightness level to another and the light reflected from the dielectric on the surface having one brightness level. 2. Apparatus as in claim 1 wherein said optical means produces light having the following characteristics in one of the paths between the transmitting means and the surface in and in the other of the paths between the surface and the receiver means: Between Transmitting Between Surface and Means and Surface (or Receiving Means (or Surfaced and Receiving Means) Transmitting Means & Surface CW only [CW, LP] [CW, UP] [CW, CCW] CW, LP CW CW, UP CW CW, CCW CW or CCW LP LP+, LP LP+, LP LP In the table above the following abbreviations are used; CW Clockwise polarization (Right handed) CCW Counter Clockwise polarization (Left hand ed) LP Linear polarization LP+ Linear polarizer aligned with LP LP Linear polarizer at blocking angle to LP > UP Unpolarized Alternating states are separated by commas. Equivalent sets of alternating states are isolated by square brackets. In any row, CW and CCW may be interchanged. In any row CW may be replaced by RH (right hand) and CCW by LH (left hand) . 3. Apparatus as in claim 1 wherein said optical means in an isolator state comprises a linear polarizer for receiving light and a quarter wave retarder plate for circu larly polarizing the light received from the linear polariz er. 4. Apparatus as in claim 1 wherein said optical means in a nonisolator state comprises a quarter wave retarder plate for circularly polarizing the light reflected from the surface and a linear polarizer for receiving the light from the retarder plate. 5. Apparatus as in claim 3 wherein said optical means in a nonisolator state further comprises a quarter wave retarder plate for circularly polarizing the light reflected from the surface and a linear polarizer for re ceiving the light from the retarder plate. 6. Apparatus as in claim 1 wherein said trans mitting means produces linearly polarized light, said opti cal means in a nonisolating state comprises a linear polar izer aligned to pass the reflected linearly polarized light and in an isolating state aligned to block the reflected linearly polarized light. 7. Apparatus as in claim 1 wherein said trans mitting means and a part of said optical means produces circularly polarized light of one hand to impinge on said surface and which is reflected as circularly polarized light of the opposite hand from a specularly reflective part of said surface and as unpolarized light from the dielectric, and another part of said optical means associated with said receiver means alternately passes and blocks the reflected circularly polarized light and passes the unpolarized light. > 8. Apparatus as in claim 7 wherein said another part of said optical means comprises a quarter wave retarder plate for receiving the reflected light and a linear polar izer for receiving the light from said retarder plate; and means for alternating the orientation about the optical axis of one of said quarter wave plate and linear polarizer relative to the light it receives. 9_ Apparatus as in claim 7 wherein said means for alternating comprises means for rotating said quarter wave retarder plate. 10. Apparatus as in claim 8 where said means for alternating comprises means for rotating said linear polar izer. 11. Apparatus as in claim 1 wherein said trans mitting means and a part of said optical means produces linearly polarized light to impinge on said surface and which is reflected as linearly polarized light from a specularly reflective part of said surface and as unpolar ized light from the dielectric, and another part of said optical means associated with said receiver means alternate ly passes and blocks the reflected linearly polarized light and passes the unpolarized light. 12. Apparatus as in claim 1 wherein said trans mitting means and one part of said optical means alternately produces circularly polarized light of opposite hand and another part of said optical means associated with said receiver means is a circular polarizer. 13. Apparatus as in claim 12 wherein said trans mitter means comprises a pair of lamps, and means for alternately energizing the lamps of said pair of lamps, said one part of said optical means co prising a circular polarizer of opposite hand for each of said lamps. 14. Apparatus as in claim 13 wherein said receiv er means comprises a video camera for producing an image of the received reflective light from each lamp, said another part of said optical means comprising a circular polarizer associated with said video camera. > 15. Apparatus as in claim 14 further comprising means for storing each of the images produced by said video camera. 16. Apparatus as in claim 14 further comprising means for displaying said images. 17. Apparatus as in claim 15 further comprising means for comparing the images produced by said video cam era. 18. Apparatus as in claim 1 wherein said trans mitter means comprises: a light source which is sequentially strobed for periods of time to produce light, said one part of said optical means is associated with said light source and produces circularly polarized light of one hand, and said another part of said optical means is associated with said receiver means and means for switching said another part between said isolating and nonisolating states. 19. Apparatus as in claim 18 wherein said another part of said optical means is in an optical path of said receiver means and is rotated on an axis transverse to its optical axis. 20. Apparatus as in claim 18 wherein said another part of said optical means comprises a housing having there in a linear polarizer and a quarter wave retarder plate parallel to each other, and means for rotating said housing. 21. Apparatus as in claim 19 wherein said receiv er means comprises a video camera. 22. Apparatus as in claim 21 further comprising means for storing each of the images produced by said video camera. 23. Apparatus as in claim 22 further comprising means for displaying said images. 24. Apparatus as in claim 23 further comprising means for comparing the images produced by said video cam era. 25. Apparatus as in claim 20 wherein said light source is a laser. > 26. Apparatus as in claim 1 wherein said trans mitting means produces circularly polarized light to the surface, said optical means being in the receiving path and comprising a retarder for changing reflected circularly polarized light to linearly polarized light, a linear polar izer, a rotator for rotating the phase of the linearly polarized light to be transmitted to said linear polarizer between blocking and transmitting phases, and a detector means for receiving the light from said linear polarizer. 27. Apparatus as in claim 1 wherein said trans mitting means and one part of said optical means produce polarized light to impinge on said surface, said receiver means comprises a pair of viewing means, and another part of said optical means comprises a first optical means associat ed with one viewing means for passing polarized light from a specularly reflective part of the surface to said one view ing means and a second optical means associated with the other viewing means for blocking polarized light from a specular part of the surface from reaching said other view ing means. 28. Apparatus as in claim 27 wherein said trans mitting means and said one part of said optical means pro duce linearly polarized light, and said first and second optical means respectively pass and block the reflected linearly polarized light from a specular part of said sur face. 29. Apparatus as in claim 27 wherein said trans mitting means and said one part of said optical means pro duce circularly polarized light, and said first and second optical means comprise circular polarizers of opposite hand which respectively pass and block the reflected circularly polarized light from a specular part of said surface. 30. A system for detecting the presence of a dielectric such as ice or snow on a metal surface comprising means for illuminating the surface with light and means for detecting the light reflected from the metal surface and the dielectric as being of two different intensities one repre > sentative of metal without the dielectric and the other of the metal with the dielectric. 31. Apparatus as in claim 30 wherein one of said illuminating and said receiving mean comprising means for alternately viewing the reflective part of the metal surface in optical isolator and nonisolator states. 32. Apparatus as in claim 1 wherein said means for receiving comprises video means having three receptive areas per pixel, to said optical means in the light path between the surface and the receiving means comprises a linear polarizer and a patterned retarder plate adjacent to said linear polarizer formed to selectively provide a refer ence thickness for one receptive area for each pixel, a thickness onequarter wave thicker than said reference thickness for a second receptive area for each pixel, and a thickness onequarter wave thinner than said reference thickness for a third receptive area for each pixel, said linear polarization axis aligned at 45 degrees to the re tarder plate "slow" axis. 33. Apparatus for detecting the presence of a dielectric substance such as ice or snow on a surface which specularly reflects light comprising: means for transmitting light over a path to said surface; means for receiving the light reflected from said surface substantially centrally located relative to said means for transmitting, and optical means in the light path between one of said transmitting means, said receiving means and the surface alternating between an optical isolator and an optical nonisolator state, the reflected light from a specularly reflecting portion of the surface reaching said means for receiving correspondingly alternating from one brightness level to another and the light reflected from a direlectric on the surface having substantially on brightness level. 34. Apparatus as in claim 33 wherein said optical means produces light having the following characteristics in > one of the paths between the transmitting means and the surface or in the other of the paths between the surface and the receiver means : Between Transmitting Between Surface and Means and Surface (or Receiving Means (or Surface and Receiving Means) Transmitting Means & Sur face) CW only [CW, LP] [CW, UP] [CW, CCW] CW, LP CW CW, UP CW CW, CCW CW or CCW LP LP+, LP LP+, LP LP In the table above the following abbreviations are used; CW Clockwise polarization (Right handed) CCW Counter Clockwise polarization (Left hand ed) LP Linear polarization LP+ Linear polarization aligned with LP LP Linear polarizer at blocking angle to LP UP Unpolarized Alternating states are separated by commas. Equivalent sets of alternating states are isolated by square brackets. In any row, CW and CCW may be interchanged. In any row CW may be replaced by RH (right hand) and CCW by LH (left hand) . 35. Apparatus as in claim 33 wherein said means for transmitting light comprises two independently con trolled light sources; a portion of said optical means being located in a first portion of said light path between one said light source and the surface comprises a first linear polarizer for receiving light and a quarter wave retarder plate for circularly polarizing the light received from the first linear polarizer, and > another portion said optical means being located in a second portion of said light path between the other said light source and the surface comprises a second linear polarizer for receiving light and producing linearly polarized light whose polarization is orthogonal to the light from said first linear polarizer, said quarter wave retarder plate circularly polarizing the light received from the second linear polarizer. 36. Apparatus as in claim 33 wherein said optical means is located in the light path between the surface and the receiving means in an isolator state comprises a quarter wave retarder plate for linearly polarizing circularly polarized light reflected from the surface and a linear polarizer for receiving the light from the retarder plate. 37. Apparatus as in claim 34 wherein said optical means is located in the light path between the surface and the receiving means in an isolator state further comprises a quarter wave retarder plate for linearly polarizing circu larly polarized light reflected from the surface and a linear polarizer for receiving the light from the retarder plate. 38. Apparatus as in claim 33 wherein said optical means in the light path between said transmitting means and the surface in an isolator state comprises a linear polariz er for receiving light and a quarter wave retarder plate for circularly polarizing the light received from the linear polarizer. 39. Apparatus as in claim 33 wherein said optical means in the light path between the surface and the receiv ing means comprises: a quarter wave retarder plate for linearly polarizing circularly polarized light of one hand reflected from the surface, a quarter wave retarder plate for linearly polarizing circularly polarized light of the other hand re fleeted from the surface, a linear polarizer for receiving the light from said retarder plates, and > means for alternately positioning said quar ter wave plates into said light path to produce optical isolator and nonisolator states. 40. Apparatus as in claim 33 wherein said optical means in the light path between the surface and the receiv ing means comprises: a quarter wave retarder plate for linearly polarizing circularly polarized light reflected from the surface, a first linear polarizer for receiving the linearly polarized light from the retarder plate, a second linear polarizer with polarization axis orthogonal to said first linear polarizer for receiving the light from the retarder plate, and means for alternately positioning said first and second linear polarizers into said light path to produce optical isolator and nonisolator states. 41. Apparatus as in claim 33 wherein said trans mitting means produces linearly polarized light, said opti cal means in a nonisolating state comprises a linear polar izer aligned to pass nearly polarized light reflected from the surface and in an isolating state aligned to block reflected from the surface linearly polarized light. 42. Apparatus as in claim 33 wherein said means for transmitting light comprises two independently con trolled light sources, and said optical means is located in the light path between said light sources and the surface and comprises a polarizing prism combiner for receiving light from both said light sources and producing a coaxial beam with orthogonal linear polarizations. 43. Apparatus as in claim 42 further comprising an optical absorber for absorbing the light not transmitted by said prism. 44. Apparatus as in claim 43 wherein said optical means further comprises two linear polarizers, each said polarizer for receiving light from one of said light sources and aligned to produce said orthogonal linear polarizations. > 45. Apparatus as in claim 44 further comprising an optical absorber for absorbing the light not transmitted by said prism. 46. Apparatus as in claim 42 further comprising a quarter wave retarder for circularly polarizing the light received from the prism. 47. Apparatus as in claim 44 further comprising an optical absorber for absorbing the light not transmitted by said prism. 48. Apparatus as in claim 33 wherein said means for transmitting beam of light scanned over said surface. 49. Apparatus as in claim 48 wherein said means for transmitting comprises a galvanometric scanner for scanning in one direction, said scanner mounted on a tilting mechanism for scanning orthogonally to said one direction. 50. Apparatus as in claim 49 further comprising means for combining the beam of light from said transmitting means and the light reflected from said surface received by said means for receiving. 51. Apparatus as in claim 50 wherein said beam combining means is a reflector. 52. Apparatus as in claim 50 wherein said trans mitting means comprises two light sources, and said optical means in the light path between said light sources and the surface comprises a polarizing prism combiner for receiving light from both said light sources and producing a coaxial beam with orthogonal linear polarizations, and a quarter wave retarder plate for circularly polarizing the light received from the prism. 53. Apparatus as in claim 52 further comprising an optical absorber to absorb light not transmitted by said prism. 54. Apparatus as in claim 52 wherein said receiv ing means comprises a photodiode, and said optical means in the light path between the surface and said receiving means compromises lens means for receiving the reflected light and focusing upon said receiving means, a quarter wave retarder plate for linearly polarizing circularly polarized > light reflected from the surface and a linear polarizer for receiving the light from the retarder plate. 55. Apparatus as in claim 52 wherein said receiv ing means comprises two avalanche photodiodes, and said optical means in the light path between the surface and said receiving means comprises lens means for receiving the reflected light and focusing upon said receiving means, a quarter wave retarder plate for linearly polarizing circu larly polarized light reflected from the surface and a polarizing prism separator for receiving the light from the retarder plate and directing linearly polarized light of one direction to one said photodiode and directing linearly polarized light of a direction orthogonal to said one direc tion to the other said photodiode. 56. Apparatus as in claim 55 wherein one said light source is omitted and said polarizing prism combiner is replaced by a linear polarizer. 57. Apparatus as in claim 54 wherein the two said quarter wave retarder plates in said optical means in the light path between said light sources and said receiving means are omitted. 58. Apparatus as in claim 55 wherein the two said quarter wave retarder plates in said optical means in the light path between said light sources and said receiving means are omitted. 59. Apparatus as in claim 56 wherein the two said quarter wave retarder plates in said optical means in the light path between said light source and said receiving means are omitted. 60. Apparatus as in claim 33 wherein said means for transmitting light provides said light at more than one intensity. 61. Apparatus as in claim 48 wherein said means for transmitting said beam of light provides said light at more than one intensity. 62. Apparatus as in claim 33 wherein said optical means in the light path between the surface and the receiv ing means provides an adjustable aperture.
2. 63 Apparatus as in claim 48 wherein said optical means in the light path between the surface and the receiv ing means provides an adjustable aperture.
3. Apparatus as in claim 48 further comprising a logarithmic amplifier for reducing the dynamic range of the output of said means for receiving.
4. Apparatus as in claim 34 where the output of said means for receiving produces an output to quantize each resolvable element of received reflected light on said surface into 3 categories: 1. Clear when the ratio of the output of said means for receiving with the optical means in said nonisolator state to the output with the optical means in said isolator state exceeds a predeter mined value determined as a function of the output of said means for receiving with the optical means in said nonisolator state and the output of said means for receiving when said means for transmit ting light is turned off, 2. Ice when not satisfying the requirements for "clear" and the output of said means for re ceiving with the optical means in said isolator state exceeds a given value, and 3. NonIce when not satisfying the require ments of "clear" and "ice".
5. Apparatus as in claim 48 wherein said means for receiving produces an output to quantize each resolvable element of received reflected light on said surface into 3 categories: l. Clear when the ratio of the output of said means for receiving with the optical means in said nonisolator state to the output with the optical means in said isolator state exceeds a predeter mined value determined as a function of the output of said means for receiving with the optical means in said nonisolator state and the output of said means for receiving when said means for transmit ting light is turned off, 2 Ice when not satisfying the requirements for "clear" and the output of said means for re ceiving with the optical means in said isolator state exceeds a given value, and.
6. NonIce when not satisfying the require ments of "clear" and "ice". 67. Apparatus as in claim 65 wherein said means for transmitting light provides light at more than one intensity and a quantized output is generated for each resolvable element on said surface as follows: 1. Clear when quantized to "clear" for any level of intensity, 2. Ice when not quantized to "clear" for any level of intensity and quantized to "ice" for the highest level of intensity, and.
7. Nonice when not satisfying conditions for "clear" or "ice". 68. Apparatus as in claim 60 wherein said means for transmitting light provides light at more than one intensity and a quantized output is generated for each resolvable element on said surface as follows: l. Clear when quantized to "clear" for any level of intensity. 2. Ice when not quantized to "clear" for any level of intensity and quantized to "ice" for the highest level of intensity.
8. Nonice when not satisfying conditions for "clear" or "ice". 69. Apparatus as in claim 65 wherein said means for receiving quantizes each said resolvable element of a continuous group of elements received at a significantly greater time delay relative to another continuous group of elements received is quantized into the "nonice" category. 70. Apparatus as in claim 66 wherein said means for receiving quantizes each said resolvable element of a continuous group of elements received at a significantly greater time delay relative to another continuous group of elements received is quantized into the "nonice" category.
9. 71 A system as in claim 30 wherein said metal surface is coated with a reflective paint. ,.
Description:
SYSTEM FOR DETECTING ICE ON METAL SURFACES

Related Application

This application is a continuation-i -part of application Serial No. 07/963,480 filed October 20, 1992 which is assigned to the same assignee.

Background of the Invention

Current airport aviation practices depend on the use of de-icing fluid to remove ice and prevent its future build-up for time periods of 5-10 minutes. Verification that wing and other aerodynamic or control surfaces are ice free is done visually, often under difficult viewing condi¬ tions. Occasionally significant ice build-ups are not noticed, with tragic results. Responsibility for detecting such ice rests with the aircraft crew who rely on visual viewing, perhaps supplemented with an ordinary flashlight. Obviously, a need exists for a system which is capable of accurately and easily determining the presence of ice on an aircraft wing.

Summary of the Invention

Metallic surfaces and dielectric surfaces behave differently when illuminated with light, particularly with respect to their polarization properties. One of the stron- gest differences and most easily observable is the property of metals to reverse the rotational direction of circularly polarized light. For example, the specular reflection of right handed (clockwise looking towards the source) circu¬ larly polarized light from a metal surface changes it to left handed (counterclockwise) polarization and vice versa. This effect is used in the construction of optical isolators which permit light to initially pass through the isolator but prevent specularly reflected light from returning through the isolator back to the source. The optical isola-

(P*

tor is a circular polarizer that is usually implemented from a linear polarizer and a quarter wave retarder plate that has its fast and slow axes located 45 * from the polarization axis of the polarizer. The polarizer must precede the retarder in the light path.

When a metallic surface (or surface painted with a metallic paint) , such as the wing of an aircraft, is illumi¬ nated with circularly polarized light (which may be generat¬ ed by passing unpolarized light through a circular polariz- er) and the reflected energy viewed through the same circu¬ lar polarizer, the resulting image is extremely dim since the circular polarizer is performing as an isolator with respect to the specular reflection from the metal surface. Other types of surfaces (birefringent, certain dielectric, matte, etc.) viewed through the same circular polarizer maintain their normal brightness because upon reflection they destroy the circular polarization. If the circular polarizer is flipped (reversed) so that the retarder pre¬ cedes the polarizer, it no longer acts as an isolator for the illuminating beam and the metallic surface's image will now be viewed of normal (bright) intensity.

Most non-metallic and painted or matte surfaces illuminated with circularly polarized light and viewed through the same circular polarizer will maintain their normal intensity. Such surfaces, as well as a coat of ice on the metal, whether matte white due to a snow covering or crystal clear due to even freezing will destroy the circular polarization of the reflected light and therefore take on the depolarizing property of a matte painted surface with respect to the optical isolator. A transparent dielectric over metal depolarizes circularly polarized light passing through it if it has numerous internal point scatterers or is birefringent. Ice has this characteristic. Thus, circu¬ larly polarized light reflected from a painted surface, snow, ice, or even transparent ice over metal will be depo¬ larized and will not be affected by the isolator.

Therefore, the image of a clear metal surface that is ice-free will alternate between dark and bright when

alternately viewed through an isolator and non-isolator structure, respectively. Apparatus other than the combina¬ tion of optical isolators and non-isolators can produce the same effect. Any ice or snow covering the metal surface will cause the image to maintain the same brightness regard¬ less of whether it is viewed through an isolator or non- isolator structure or equivalent structures.

The present invention provides various arrange¬ ments for inspecting a metal surface for the presence of ice which compares views of the surface in an optical isolating and non-isolating manner. Making such comparisons in an alternating manner results in the metal surface producing a blinking, on-off, viewing of the reflected light and the ice producing a steady level of illumination. In accordance with the invention, various embodi¬ ments are provided for inspecting a metallic surface in which there is a comparison or switching between an optical isolator structure and non-isolator structure. In one embodiment, switching is implemented by switching the light illuminating the metal surface between circularly polarized and non-circularly polarized light while observing through a circular polarizing filter of the same hand, e.g., C or CCW, as required to complete the isolator. In another embodiment, the light illuminating the surface may be kept circularly polarized but viewed alternately through a circu¬ lar polarizer of the same hand and a non-circular polarizing element having the same optical attenuation. This is most easily accomplished by viewing through the same type of circular polarizer flipped over (reflected light enters the polarizing element first) to keep it from acting as the circular to linear polarizing element of an isolator while simultaneously maintaining the slight light attenuation of its linear polarizer element.

Another embodiment maintains the illumination in a circularly polarized state and alternately views the scene through right handed and left handed circular polarizers which will alternately change between the isolating and non- isolating states. A non-isolating state may also be

achieved by rotating either the receiver or transmitter quarter wave retarder plate forming a part of the polarizer by 45° . This aligns the slow or fast axis of the retarder with its polarizer. The effect is that, if done at the transmitter, linearly polarized light passing through the quarter wave plate remains linearly polarized and if done at the receiver, circularly polarized light (which passes through the retarder plate first) emerges linearly polarized at 45° to the original direction - it can then pass through the linear polarizer with just slight attenuation.

Rotation of either the transmitter or receiver quarter wave retarder by 90° from the position in which it serves to operate as an isolator also changes the state to non-isolating because the specularly reflected circularly polarized wave is then exactly aligned with the receiver polarizer as it emerges in the linearly polarized form from the receiver's quarter wave retarder. Isolating and non- isolating states may also be achieved by various combina¬ tions of crossed and aligned linear polarizers, respective- ly.

Since the reflected light from a specular surface is highly directional, it is beneficial to minimize any change in illumination angle when changing from isolator to non-isolator state. Otherwise a change in reflected light intensity caused by a change in illumination angle may be interpreted as caused by the isolator/non-isolator effect and an erroneous decision made. The high directionality of specular reflection also introduces the need to accommodate a large dynamic range of received light intensities. If not properly handled, erroneous decisions could be made as a result of saturation due to high light levels within the receiver.

Background light such as sunlight must be removed from influencing the final clear/ice surface decision or the system will be limited to operation in low light levels.

Light reflected from surfaces other than the aircraft wing, such as the ground, when viewing downward on the wing must

also be removed in order to provide an unmistakable image of the wing and any patches of ice on it.

Specular surfaces that can be viewed close to the surface normal provide a relatively high isolator/non-isola- tor ratio and therefore a clear demarcation between the two light levels. However, as the surface is viewed at angles away from normal to the surface, which is necessary for a system viewing from a fixed location, the ratio drops and the reflected light effect becomes closer to that produced by reflection from ice, making it more difficult to reliably distinguish between the two.

Object of the Invention

It is therefore an object of the present invention to provide an apparatus for detecting the presence of a depolarizing dielectric material, such as ice or snow, on a metal specular reflecting surface.

A further object is to provide a system for de¬ tecting ice and/or snow on the metal (or metallic painted) wing of an aircraft.

An additional object is to provide a system for detecting ice and/or snow on a metal (or metallic painted) surface which is specularly reflective to light using circu¬ larly or linearly polarized light. Yet another object is to provide a system for detecting ice or snow on a metal or metallic painted surface in which optical means are used to produce an on-off light blinking response for a metal surface and a steady light response for any part of the surface covered with ice or snow.

A further object is to provide a process and apparatus to overcome system limitations brought about by ambient light, the large dynamic range of light intensities and the poor response of metal surfaces viewed away from a direction normal to the surface.

Other objects and advantages of the present inven¬ tion will become more apparent upon reference to the follow¬ ing specification and annexed drawings.

" * * *

Brief Description of the Drawings

Figure 1A is an optical schematic of a circular polarizer with the linear polarizer facing the illumination source so that the polarizer acts as an optical isolator. Figure IB is an optical schematic of a circular polarizer with the quarter wave plate facing the illumina¬ tion source so that it passes specularly reflected light, i.e., it is a non-isolator;

Figure 2 is an optical schematic of two circular polarizers, one in the transmit path and one in the detec¬ tion path, that together form an optical isolator;

Figure 3A is a schematic view of an ice detection apparatus based on direct visual observation using two spot¬ light illuminators, one polarized and one not; Figure 3B is a schematic view of an ice detection apparatus based on direct visual observation which uses one circularly polarized light source;

Figure 3C is a detail of the Figure 3B apparatus for switching the polarizer between isolating and non-iso- lating states in the detection path;

Figure 4A is a schematic diagram of a video based ice detection system suitable for use with high background illumination levels which employs two strobed light sources;

Figure 4B is a schematic diagram of a video based ice detecting system employing one laser based strobed light source suitable for use with high background illumination levels;

Figure 5A is a schematic view of the device used in Figure 4B to switch the polarizer from an isolating to a non-isolating state in the detection path;

Figure 5B is a plan view of the motor, polarizer and encoder assembly used with the apparatus of Fig. 5A;

Figure 5C is a schematic view of the photo inter¬ rupter device used in the encoder assembly of Figure 5B; - Figure 6 is an optical schematic diagram of the laser light source of the system of Figure 4B;

Figure 7 is a schematic of another embodiment of the invention which utilizes synchronous detection;

• -

Figure 8A is a schematic view of an embodiment which uses two video cameras and a beam splitter device;

Figure 8B is a schematic view of the optical path of Figure A using two mirrors to replace the beam splitter device of Figure 8A;

Figure 9A is a schematic view of a polarization sensitive camera based upon a variation of color camera technology which is particularly suitable for use in the receive path; Figure 9B is a section view of details of the polarization sensitive camera;

Figure 10a is a side view of an ice detection system light projector/receiver viewing an aircraft wing;

Figure 10b is a front view of the light projector /receiver;

Figure 10c is a front detail view of one light source;

Figure lOd is a schematic view of the receiver portion of the ice detection system; Figure lOe is an alternate version of the receiver portion of the ice detection system using alternating quar¬ ter wave retarder plates;

Figure lOf is an alternate version of the receiver portion of the ice detection system using alternating linear polarizer plates;

Figure 11 schematically depicts the view coverage of the system;

Figure 12 is a schematic of an alternative light source producing coaxial beams for isolator and non-isolator states;

Figure 13 is a schematic of a scanning narrow beam ice detection system;

Figure 14 depicts the time relationship of waveforms within the system as a function of surface dis- tance;

Figure 15 is a graph depicting the receiver input ratio as a function of angle from surface normal for surfac¬ es with different properties;

Figure 16A, 16B and 16C together comprise a flow diagram of the process for categorizing the measured values as clear, ice, non-ice;

Figure 17a shows the waveforms for an analog system for determining leading edge delay; and

Figure 17b shows the waveforms for a digital system for determining leading edge delay.

Detailed Description of the Invention Figure 1A illustrates the operation of a circular polarizer used as an isolator. Light is emitted from an unpolarized source 13, which preferably is as close to mono¬ chromatic as possible. The light is shown as unpolarized by the arrows in two orthogonal directions along line 20, the path the light is following. The unpolarized light passes through a linear polarizer 11 which has a vertical polariza¬ tion axis. The light passing through linear polarizer 11 takes path 21, along this path illustrated as vertical polarization by the double arrow. The vertically polarized light at 21 passes through a quarter wave retarder plate 12. The retarder 12 is a plate made from birefringent material, such as mica or crystal quartz. Its purpose is to change linearly polarized light from polarizer 11 into circularly polarized light. Any ray incident normal to the retarder plate 12 can be thought of as two rays, one polarized parallel to the parent crystal's optic axis (e-ray) and the other perpendicular (o- ray) . The e-ray and o-ray travel through the plate 12 at different speeds due to the different refractive indices. The plate 12 is said to have a "fast" and a "slow" axis. The quarter wave retarder plate 12 has its slow and fast axes both at 45° relative to the vertical axis of the linear polarizer 11 so that the emerging circular polar¬ ized light from plate 12 along path 22 is rotating in a CCW direction as viewed facing the light source from a reflect¬ ing surface 14. A metallic surface, which is a specular reflector, and a dielectric surface, i.e., ice or snow, behave differently when illuminated with light, particularly

*•_ J !

with respect to their polarization properties. A strong and easily observable difference is the ability of a metal to reverse the rotational direction of incident circularly polarized light. The specular reflection of right-handed (CW) circularly polarized light from a metal surface changes into left-handed (CCW) polarization and vice versa.

This effect is used in the construction of optical isolators which permit light to initially pass through the isolator but prevent such light when specularly reflected from returning through the isolator back to the light source. When the optical isolator is a circular polarizer it is usually implemented from a linear polarizer and a quarter wave retarder plate that has its fast and slow axes located 45° from the polarization axis of the polarizer. In Fig. 1A, surface 14, which is a specular sur¬ face, reflects the incident circular polarized light back along path 23. The reflected light continues to rotate as viewed from the surface 14 in the CCW direction but has now changed "hand", in terms of right hand and left hand, be- cause it is rotating in the same direction with its direc¬ tion of travel changed.

The reflected light on path 23 passes through the quarter wave retarder 12 and emerges no longer circularly polarized but linearly polarized in the horizontal direc- tion, which is shown along ray path segment 24. Because the light ray 24 is horizontally polarized it is not passed by the (vertical) linear polarizer 11. Therefore, none of the specularly reflected light gets through to path segment 25 to enter the eye 26, which is shown near the location of the light source 13. Thus, the quarter wave retarder plate 12 acts as an optical isolator. That is, light from the source 13 is passed through the circular polarizer and reflected by the specular surface 14 but cannot pass through the circular polarizer back in the other direction and so is blocked before it gets to the eye.

Figure IB shows the same quarter wave plate and linear polarizer combination used, but the sequence of the elements is reversed. Here, the quarter wave retarder plate

12 is facing the illumination source 13 and the linear polarizer 11 is facing the output side towards the reflect¬ ing surface 14. The light rays now emerge from source 13 in an unpolarized form along ray path 20 and pass through the quarter wave plate 12. However, because the light is not polarized the quarter wave plate 12 does not change any polarization properties. The light then passes through the linear polarizer 11 and becomes vertically polarized along' ray path 22. Surface 14 specularly reflects with the same polarization the vertically polarized light which travels along ray path 23 back towards the linear polarizer 11 with the same polarization. The light now enters the quarter wave retarder plate 12. Because the light entering plate 12 is polarized in the vertical direction, it emerges from the quarter wave plate circularly polarized. However, this is of no consequence to the eye 26, so the eye sees the light that has been reflected from the surface 14. Thus, in this case with the light first entering the quarter wave plate 12 and then passing through the linear polarizer 11 and being specularly reflected back to the eye through the linear polarizer and the quarter wave plate, there is little loss in the light intensity.

As can be seen in the comparison of Figs. 1A and IB, light from the same source 13 reflected from the specu¬ lar reflection surface 14 is viewed by the eye 26 either dim or bright depending upon the location of the quarter wave retarder plate 12 relative to the linear polarizer 11. That is, Fig. 1A effectively is an optical isolator while Fig. IB is a non-isolator.

Figure 2 shows the same implementation of a circu¬ lar polarizer as in Figure 1A, with the receive path and the transmit path each having their own circular polarizers. Both circular polarizers are in the same order. That is, both linear polari-zers 11a and lib are on the left, one adjacent to the light source 13 and the other the eye 26, and both quarter wave retarders 12a, 12b are on the right adjacent to the reflective surface. Thus, as shown, the

light from lamp 13 enters the linear polarizer 11a, exits vertically polarized, passes through the quarter wave plate 12a and emerges rotating CCW as viewed from the specular reflecting surface 14. The light reflects off the surface 14 still polarized rotating CCW as viewed from surface 14 and passes through the circular polarizer 12b in the return direction path to enter quarter wave plate 12b, from which it exits horizontally polarized to the vertical linear polarizer lib which blocks the light. Linear polarizer lib in the reception leg is distinct and separate from the linear (vertical) polarizer 11a that was used in the trans¬ mit leg. Because the polarization of the light ray along path 24 is horizontal, the light does not pass through the linear polarizer lib and cannot enter the eye 26. When a metallic surface, such as the wing of an aircraft, is illuminated with circularly polarized light produced by the device of Fig. 1A and the reflected energy viewed through the same circular polarizer the resulting image is extremely dim since the circular polarizer is performing as an isolator with respect to the specular reflection of the circularly polarized light (of opposite hand) from the metallic surface.

A painted portion (non-specular) of the surface illuminated with circularly polarized light does not reflect light in a polarized form. Instead, it destroys the circu¬ lar polarization and makes the reflected light unpolarized. Thus, the unpolarized light reflected from a painted surface portion when viewed through the same circular polarizer of Fig. 1A will maintain its normal intensity. The same holds true for circular polarized light reflected from a wing covered by ice or snow. However, other common harmless substances such as water or de-icing fluid that may be on the wing do not destroy the circular polarization of the reflected light. As explained with respect to Fig. IB, the compo¬ nents of the circular polarizer of Fig. 1A are flipped (rotated) such that the retarder plate 12 precedes the linear polarizer 11 with respect to the source of light 13,

so it no longer acts as a circular polarizer to an illumi¬ nating beam. Accordingly, the reflection of circular polar¬ ized light from the metal surface will pass back to the eye and will be of normal (bright) intensity. The image inten- sity of such light reflected from a painted or dielectric (non-specular) surface also will be unchanged as in the previous case.

When a metallic surface is alternately illuminated and viewed by the isolator and non-isolator devices of Figs. 1A and IB, the return images at the eye 26 will alternate between dark and bright. A painted or dielectric non-specu¬ lar surface will remain uniformly bright to the alternation since the light reflected from the painted or dielectric surface is not polarized and will not be isolated. Assuming that a metallic surface has a patch of ice thereon or is coated with ice, the ice being either matte white due to snow covering or crystal clear due to even freezing, this will destroy the circular polarization of the reflected light and therefore take on the property of a matte painted surface with respect to the optical isola¬ tor. That is, referring to Fig. 1A, if there is ice on any portion of the specular surface 14, then the circularly polarized light along path 22 impinging upon such portion of the surface will not have its polarization reversed. In- stead, it will have the effect of a painted surface so that the returned light will be non-polarized and will pass to the eye, i.e., the returned image will be bright.

Accordingly, upon alternately illuminating and viewing an ice-free metallic surface 14 with the circular polarizer-isolator of Fig. 1A and the non-isolator of Fig. IB, the return viewed by the eye 26 will alternate between dark and bright respectively. Any ice or snow covering a portion of the metal surface 14 will cause that portion of the image to maintain the same brightness regardless of whether it is viewed through an isolator or non-isolator structure upon such alternate illumination and viewing.

Switching between an isolator structure, e.g., Fig. 1A, and non-isolator structure, e.g., Fig. IB, may be

implemented by switching the light illuminating the metallic surface between circularly polarized and non-circularly polarized light while observing through a circular polariz¬ ing filter of the same hand, e.g., CW or CCW, as required to complete the isolator. As an alternative, the light illumi¬ nating the metallic surface may be kept circularly polarized but viewed alternately through a circular polarizer of the same hand and a non-circular polarizing element having the same optical attenuation. This is most easily accomplished by viewing through the same type of circular polarizer flipped over (reflected light enters the polarizer element first) to keep it from acting as the circular to linear polarizing element of an isolator while simultaneously maintaining the slight light attenuation of its linear polarizer element.

Another arrangement is to maintain the illumina¬ tion in a circular polarized state. Thereafter, the surface would alternately be viewed through right-handed and left- handed circular polarizers which alternately change between the isolating and non-isolating states.

A non-isolating state also may be achieved by rotating either the receiver or transmitter quarter wave retarder plate 12 by 45°. This aligns the slow or fast axis of the retarder with its polarizer. The net effect is that, if done at the transmitter, linearly polarized light passing through the quarter wave plate remains linearly polarized. If done at the receiver, circularly polarized light (which passes through the retarder plate first) emerges linearly polarized at 45° to the original direction. It can then pass through the linear polarizer to be viewed with just slight attenuation.

Rotating either the transmitter or receiver quar¬ ter wave retarder by 90° from the position in which it serves to operate as an isolator also changes the state to non-isolating because the specularly reflected circularly polarized wave is then exactly aligned with the receiver polarizer as it emerges in linearly polarized form from the receiver's quarter wave retarder.

e \

The following table illustrates the implementation that may be used when alternating either the illumination

(transmitter) or receiver (detector) polarizing elements or vice versa to change the overall path from an isolator to a non-isolator structure:

Between Transmitter Between Surface and Surface (or and Receiver (or

Surface and Receiver) Transmitter and Surface) CW only [CW, LP] [CW, UP] [CW, CCW]

CW, LP CW

CW, UP CW

CW, CCW CW or CCW LP LP+, LP-

LP, UP LP-

In the table above the following abbreviations are used; CW Clockwise polarization - (Right handed) CCW Counter Clockwise polarization - (Left hand¬ ed)

LP Linear polarization LP+ Linear polarizer aligned with LP LP- Linear polarizer at blocking angle (e.g. 90°) to LP

UP Unpolarized

Alternating states are separated by commas. Equivalent sets of alternating states are isolated by square brackets. In any row CW and CCW may be interchanged.

In any row CW may be replaced by RH (right hand) and CCW by LH (left hand) . (The columns can be interchanged), i.e., the action can be either on the transmitter or receiver leg.

The table shows that when using linear polariza¬ tion the isolating state refers to the receiving polarizer being orthogonal to the polarization of the transmitted

energy beam and the non-isolating state refers to any of the following conditions: a) non-polarized transmission b) no polarizer in receiver path c) polarizer in receiving path is approximately aligned with the polarization of the transmitted beam. Figure 3A is a schematic view of a monocular version of an ice detection system suitable for night use based on direct visual observation. The direct visual observation receiver uses a non-inverting telescope 50 with a circular polarizer 40, like the circular polarizer of Fig. 1A, at its entrance. Two spotlights 13a, 13b are used for the source of illumination, i.e., the transmitter. One spotlight 13a has a circular polarizer 30 isolator, like Fig. 1A, mounted to it. The other spotlight 13b has a neutral density filter 30a or a "same hand" circularly polarized filter mounted backwards so that the light coming through is linearly and not circularly polarized, i.e., like the non-isolator of Fig. IB. The two spotlights 13a, 13b illuminate a common overlapping area of a surface shown as the entire area or a portion of an aircraft wing 15 having a patch 16 of ice thereon. The clear (no ice) portions of the wing 15 form a specular reflecting surface such as the surface 14 of Figs 1A and IB. The wing 15 is observed by the field of view 23 of the non-inverting telescope 50. Both spotlights 13a, 13b and the non-inverting telescope 50 are mounted on a support structure 52, which in turn is mounted to a tripod or boom 54. A power supply and sequencer 51 for the lights 13a, 13b is also located on the tripod boom structure. Two outputs from the sequencer 51 are taken along wires 53a and 53b to connect with and alternately energize the lamps 13a and 13b, respectively.

The eye 26 is shown looking through the telescope 50 The field of view of the upper spotlight 13a is shown as 22a and that of the lower spotlight 13b as 22b. The region observed by the non-inverting telescope 50 is formed from the fan of rays 23 reflected back from wing 15 into the

26

telescope 50. In operation, the sequencer 51 alternates between sending a voltage to and alternately energize spotlight 13a and then spotlight 13b during corre¬ sponding time periods "a" and "b". When the voltage is applied to spotlight 13a the outgoing light is circularly polarized by polarizer 30 and the light emerges in fan 22a which illuminates the aircraft wing surface 15. The light from fan 22a reflected from the aircraft wing 15 passes back through fan 23 into the circular polarizer 40 of non-invert- ing telescope 50 where it may be viewed by the eye 26 during the interval "a" . During the period "a" an optical isolator arrangement is in place because there are two circular polarizers 30 and 40 in the path. This is shown in Fig. 2. That is, metal areas of the wing which produce a specular mirror like reflection reverse the "hand" of the incident circularly polarized light and prevent it from passing back through the isolator. Therefore, the eye 26 sees a very dark region covering the aircraft wing, except where there is ice, which is shown on area 16 of the aircraft wing and which area will show brighter to the eye through the tele¬ scope.

When spotlight 13a is turned off and spotlight 13b is turned on during period "b", the light emerging from spotlight 13b is not circular polarized. Now the reflection coming back to telescope 50 from both the areas with ice or a metal area without ice will approximately maintain their normal brightness. Thus as the sequencer 51 alternately energizes the spotlights 13a and 13b, the image at the eye 26 from any area that is metal, specular and ice free will appear to blink on and off. This will be "on" (bright) when the optical isolator is not in operation and "off" (dark) when isolation exists. However, areas that have ice will not blink and will have essentially constant brightness, because the polarized light produced during period "a" is de ~ -polarized upon impinging and being reflected from the ice or the metal under the ice.

Figures 10a-f illustrates a concentric arrangement of illuminating light sources 13 surrounding a camera 80 all

5" <^t

mounted in a supporting structure 52 forming an ice detec¬ tion system suitable for daylight use. Figure 10a shows a side view of structure 52 with light beams 22a and 22b aimed toward aircraft wing 15 with an area of ice 16. Reflected light from surfaces of wing 15 and ice 16 are collected within the viewing angle 23 of a camera 80 mounted on struc¬ ture 52. Light sources 13 contain sources of linearly polarized light that project light through quarter wave retarder windows 30a and 30b shown in side edge view in Fig. 10a to produce beams 22a, 22b. Windows 30a and 30b are tilted away from the optic axes of beams 22a and 22b respec¬ tively, in order to prevent light reflections from aircraft surfaces 15 (metal) and 16 (ice) from being redirected from windows 30a, 30b back towards surfaces 15 and 16. The combination of linearly polarized light from sources 13 and quarter wave retarder window 30a, 30b produce circularly polarized light beams 22a and 22b. Reflected light from non-ice covered surface 15 is circularly polar¬ ized whereas light reflected from ice surface 16 is sub- stantially unpolarized.

Figure 10b shows a front view of structure 52 with six light sources 13a-13f surrounding camera 80. Each light source 13 is similar in construction and has an array of linearly polarized light sources behind quarter wave retard- er window 30a. As shown in Figure 10c, each light source 13 has four similar segments 203-1 through 203-4. Each segment 203 has two assemblies 203a and 203b that have a plurality of linearly polarized light sources 103. Each light source 103 is, for example, an LED (such as AND180CRP which produc- es an 8° beam) mounted behind a linear polarization filter. The polarization axis of the filter is preferably oriented at 45° to the vertical to minimize preferential reflection of the light from the surface 15 to be illuminated. The polar¬ ization axes of the filters on assemblies 203a and 203b are mounted orthogonal to each other to provide the basis for distinguishing areas containing ice such as area 16 on surface 15.

ULE 26

Assemblies 203a and 203b are placed in close proximity to assure that the light projected from each, impinge on surface 15 at nearly equal angles of incidence to minimize differences in reflected energy and to provide the largest possible depth over which the beams of light from each assembly 203a and 203b coincide. Assemblies 203a and 203b are preferably independently adjustable in two direc¬ tions to enable alignment of the light beams from these assemblies. They can each be tilted vertically and rotated horizontally by amounts to provide overlap of the beams from light sources 13a through 13f.

Figure lOd shows camera 80 with associated quarter wave retarding plate 12 to convert circularly polarized light received within camera view angle 23 to linearly polarized light that can be blocked by linear polarizer 11 (when the polarization is orthogonal to the polarization axis) .

During operation of the system of Fig. 10, all assemblies 203a of light sources 13 are strobed to produce light circularly polarized of one hand. Then when camera 80 is ready to record the next image, all assemblies 203b of light sources 13 are strobed to produce light circularly polarized of the other hand. Specularly reflected light from the metal part of surface 15 will be rejected in one image and not in the other, creating a blinking effect.

Light reflected from ice surface 16, however, will be sub¬ stantially non-polarized and recorded in all images formed within camera 80 without a blinking effect.

The images can then be directly displayed on a TV monitor for an observer to determine if any ice is present. The images also can be processed and presented to an observ¬ er as enhanced images clearly defining areas where ice 16 is present on surface 15.

An alternate arrangement to that shown in Figs. 10a-lOd is to omit quarter wave retarding plate windows 30a and 30b from the light sources 13, and quarter wave retard¬ ing plate 12 from camera 80. Linear polarizer 11 will reject light reflected from clear portions of surface 15

when the projected polarization is orthogonal to its polar¬ ized axis and pass light from the same surface when the projected polarization is aligned to its polarization axis. A further alternative arrangement is to use a single linear polarization filter (or separate filters with their polarization axes aligned) in front of light sources 103 on assemblies 203a and 203b. Figure lOe provides a detail of camera 80 surrounded by rectangular quarter wave plate segments 12a and 12b bent into half-cylinders to form a barrel 61 that is rotated to alternate the two half-cylin¬ der segments 12a and 12b to be in front of camera 80 when light source 13 is strobed. One quarter wave plate segment 12a is arranged to accept circularly polarized light of one hand received in camera view angle 23 and convert it to linear polarization aligned to linear polarizer 11, thus allowing the light to pass to camera 80. The other quarter wave plate segment 12b is arranged to accept the same light and convert it to linear polarization orthogonal to the polarization axis of linear polarizer 11, thus preventing the light from reaching camera 80.

Using the camera of Fig. lOe, all assemblies 203a and 203b of the light sources 13 are strobed simultaneously when camera 80 is ready to record an image. Barrel 61 is rotated to synchronize the positions of segments 12a and 12b to be present alternately for sequential images. Light sources 13 project circularly polarized light of one hand on a surface 15. Light reflected from specular portions of surface 15 will be circularly polarized and produce a blink¬ ing image in camera 80 as segments 12a and 12b alternately cause the reflected light to be passed or blocked respec¬ tively. Unpolarized reflected light from ice patches 16 will pass to camera 80 when either segment 12a or 12b is present and the images will not blink. Further, since the closely spaced light source assemblies 203a and 203b are strobed for both images, there is no angular shift in the light source that could alter the reflected light intensity. This produces less light intensity variation from specular surfaces that could tend to reduce the blinking effect.

A further alternate arrangement omits the quarter wave retarding plate windows 30a and 30b from light sources 13. Figure lOf shows video camera 80 surrounded by rectan¬ gular linear polarizer segments 11a and lib bent into half- cylinders to form a barrel 61 that is rotated to alternate the two segments previously described. Segment 11 is ar¬ ranged to have its polarization axis aligned to the polar¬ ization received in camera angle 23, reflected from surface 15. Segment lib is arranged to have its polarization axis orthogonal to the same light.

Using the camera of Fig. lOf all assembles 203a and 203b of the light sources 13 are strobed simultaneously and produce linearly polarized light of one polarization when camera 80 is ready to record an image. Barrel 61 is rotated to synchronize the positions of segments 11a and lib to be present alternately for sequential images. Light reflected from specular portions of surface 15 will be linearly polarized and produce a blinking image in camera 80 as segments 11a and lib alternatively cause the reflected light to be passed or blocked respectively. Unpolarized reflected light from ice patches 16 will partially pass to camera 80 when either segment 11a or lib is present and the images will not blink.

A further alternative retains the quarter wave retarding plate windows 30a and 30b of the light sources, and adds quarter wave retarding plate 12 shown in Figure lOf. This produces circularly polarized light of one hand from light sources 13 for projection upon surface 15. Plate 12 then converts the circularly polarized light from specu- lar portions of surface 15 to linearly polarized light aligned to the polarization axis of segment 11a and orthog¬ onal to the polarization axis of segment lib. Thus a blink¬ ing image will be produced for specular portions of surface 15, but not for iced portions 16 that reflect light that is substantially unpolarized.

Figure 11 is a detail of the coverage of the ice detection system of Figures 10a-f. Light beams 22a and 22b from light sources 13 are aimed to illuminate surfaces at

distances between 25 and 26 from camera 80 and within camera view angle 23. This is done in both the vertical and hori¬ zontal planes.

Another form of light source 13 of Fig. 10 is shown in Figure 12. A lamp 13a, such as a flash lamp, has its light shaped into a beam 20a by optics. Beam 20a is linearly polarized by polarizer 11a forming linearly polar¬ ized beam 21a whose polarization is aligned to the polariza¬ tion axis of polarizing prism combiner 503. Polarizing beam combiner 503 is a polarizing beam splitter such as Melles Griot 03PBS049 which is used in reverse to form a beam combiner. Beam 21a passes through prism 503 and quarter wave retarder plate 12 to form circularly polarized beam 22. Likewise lamp 13b forms beam 20b, passes through linear polarizer lib forming linearly polarized beam 21b whose polarization is orthogonal to the polarization axis of prism 503. Beam 21b is turned 90°, passes through plate 12 and can be aligned to coincide with beam 22 from beam 21a. Since the two beams 21a, 21b coincide, any angular offset effects of the light source is avoided as the lamps 13a and 13b are alternately strobed for reception by a camera 80 as shown in Figure lOd. The principles of operation are as previously described, with lamp 13a of Fig. 12 replacing assembly 203a of Fig. 10c and lamp 13b replacing assembly 203b. Alter- nately, quarter wave plates 12 in Figure 12 and Figure lOd may be discarded to produce a system relying on linearly polarized light rather than circularly polarized light as previously described.

The linear polarizing plates 11a and lib in Figure 12 are not required if an absorber 17 is placed as shown to absorb the orthogonal polarized light of beam 21a, which will be reflected by prism 503, and the aligned polarized light of beam 21b, which will pass through prism 503.

The intensity of light reflected from a specular surface can vary by many orders of magnitude depending on the viewing angle. If camera 80 has an insufficient dynamic range, specular reflections of high intensity can cause saturation in portions of the image that can spread to

adjacent areas providing a distorted image and may obscure the blinking effect, preventing the proper system operation.

One solution to the dynamic range problem is to first form the blocking/non-blocking image pairs with a 5 normal level of light intensity and then with a lower level projected light intensity. The overall dynamic range of the system is thus increased by the amount (say 10:1) of the light reduction. If a greater dynamic range is required, a third image pair can be made at a further reduction. The

10 overall dynamic range of the system is thus increased by the product of the reductions (say 10x10 = 100:1) .

The reduction of the amount of light projected by light source 13 does not improve on the interference intro¬ duced by background light energy that may be within the

15 field of view and add to the image. A preferred method of increasing dynamic range, while at the same time reducing background interference, is to take several blocking/non- blocking image pairs where each pair is taken at a reduced camera sensitivity by reducing the aperture size or aperture

20. time of camera 80 while keeping the projected light inten¬ sity constant from light source 13. Background interference can be further reduced by filtering the light received by camera 80 with a filter transparent to the wavelengths of light projected by light source 13 and blocking background

25 light of other wavelengths. Because a sensor's response to light may be non-linear and the ratios desired are not necessarily constant, it may not be possible to merely subtract the observed value of background light. Rather the background light and the non-isolator light may be used as

30 an index (or indices) in a look-up table of predetermined values to serve as a threshold for determining if the isola¬ tor light level is an indication of the presence of ice. The method for automated processing is described below.

Figure 3B shows another ice detection apparatus

35 especially suitable for night use, which is based on direct visual observation and uses only one spotlight 13 with a circular polarizer 30, such as Fig. 1A. In Figure 3B the receiver telescope 50 has apparatus at its input for chang-

ing a circular polarizer between the isolating (Fig. 1A) and non-isolating (Fig. IB) states. Here, the illumination source 13 and the telescope 50 are mounted on a bracket 52 of a boom mount or tripod 68. A power supply 67 for lamp 13 also is mounted on the boom.

Power supply 67 supplies the power to lamp 13 along cable 66b. Lamp 13 incorporates a circular polarizer 30, such as of Fig. 1A. The field illuminated by lamp 13 is shown as 22a and encompasses an aircraft wing area 15 which has an area of ice 16. Telescope 50 has a field of view encompassing the aircraft wing, or portion of the wing, and this is shown in the ray fan 23 which enters the telescope. Telescope 50 alternates between optical isolation and non- isolation to the reflected light using a circular polarizer made of a fixed linear polarizer 11 and quarter wave retard¬ er plate 12. As shown in Fig. 3C, the quarter wave retarder plate 12 is rotated about its optical axis by drive 60.

Figure 3C is a detail showing the apparatus for rotating the quarter wave retarder plate 12. The quarter wave retarder plate is rim driven by friction drive 65 attached to a motor shaft 64 driven by a motor 63 which itself is attached to telescope housing 61. Bearings 62 between the quarter wave retarder plate 12 and the housing 61 relieve friction so that the quarter wave retarder plate may freely rotate about its optical axis. When the quarter wave plate has rotated to such a position that its slow and fast axes are at 45° to the vertical, as shown in Fig. 2, the unit acts as an optical isolator and any circularly polarized light that is specularly reflected from the air- craft wing cannot pass through the combination of the quar¬ ter wave retarder and the linear polarizer to the eye 26.

A similar end may be achieved by rotating the linear polarizer 41 via rim drive 60 and keeping the quarter wave retarder plate 42a fixed or by keeping both linear pσlarizer 41 and quarter wave retarder plate 42a fixed and rotating a half wave plate mounted between them with rim drive 60.

"

The position for optical isolation is achieved twice during two positions spaced 180° apart of each revolu¬ tion of the quarter wave retarder 12. At any other position of rotation of plate 12, there is no isolation and the circularly polarized light reflected from the various por¬ tions of the wing, both metal and ice, is free to pass through to the eye with only minimal attenuation. There¬ fore, the specularly reflective metal portion of the wing that is not covered with ice will reflect light from the illuminator 13, circularly polarized, back through the isolating mechanism 40 and this specularly reflected light will be interrupted twice per revolution and blink off completely. During the other positions of the circular polarizer retarder plate 13 rotation the light will pass through to the eye 26. Thus, the "on"-"off" blinking effect will be produced twice for each rotation of plate 12.

On the areas of the wing 15 when there is ice present, the incident circularly polarized light from lamp 13 and polarizer 30 will be de-polarized due to the surface of the ice or by passing through the ice. This de-polarized light will pass through the isolator 40 at the telescope 50 regardless of the rotational position of the quarter wave retarder plate 12. That is, even when the plate 12 is in one of its two isolating positions relative to reflected polarized light, the non-polarized light reflected from the ice will pass through to the telescope as well as when the retarder plate is in a non-isolating position.

The eye 26, which is looking through the telescope 50, is able to differentiate between the blinking effect produced by the ice free section of the wing 15 and the non- blinking effect produced by sections 16 of the wing with ice. That is, the sections of the wing covered with ice 16 will appear to have constant illumination and the ice free sections of the wing will appear to blink at a rate of twice the speed of rotation of the quarter wave plate 12.

In either of the embodiments of Figs. 3A and 3B, the apparatus can be moved to scan all parts of the wing if the field of view is not large enough.

'

Figure 4A shows an indirect viewing video-based ice detecting system that employs two strobe lamp spotlights and is suitable for use with high background illumination levels. The system of Fig. 4A is similar to that of Fig.

3A in that it employs two strobe lamps 93a, 93b. These lamps are of the type which produce a high intensity output, for example a xenon lamp, for a short time period. Here, both strobe lamp 93a and 93b have circular polarizers, such as in Fig. 1A, attached. One is a right handed circular polarizer 30a and the other a left handed circular polarizer 30b. The strobe lamps 93a and 93b are used in conjunction with a conventional video camera 80 with a lens 81 having a right handed circular polarizer 40 at its input. The analog signals of the image produced by the video camera, which observes the scene illuminated by the strobe lamps 93a, 93b, are sent to a conventional frame grabber 70. The frame grabber 70 converts the analog video signal from camera 80 to digital form and stores them in a digital memory buffer 70a. Pulse generator 75 is used to initiate the strobing of the lights and the grabbing of a single isolated frame by the frame grabber from the video camera.

The system also preferably has a digital to analog converter and sync generator so that the image stored in the buffer 70a can be sent from the frame grabber video output to a video monitor and/or VCR 72 along cable 71. The video monitor and the video cassette recorder (VCR) are commer¬ cially available. As an alternative, the video monitor may have a disk recorder which is also commercially available. The frame grabber may be purchased with additional memory attached and a computer as part of one single image proces¬ sor unit. This portion is shown as 70A. Frame grabber 70 and its memory, plus computer CPU 90A may be bought commer- cially as the Cognex 4400.

A flip flop 85 alternates between states on every strobe pulse produced by pulse generator 75. This allows selectively gating a strobe pulse to either lamp 93a or 93b

so that they are illuminated alternately. When a pulse trigger input is received by the frame grabber 70 from pulse generator 75 output 76, a camera synchronized strobe pulse is generated which is fed from the frame grabber output 74 to the flip flop 85. The strobe pulse toggles flip flop 85 and is also gated through one of two AND gates 89a and 89b. When the flip flop 85 is in one state the strobe out of the frame grabber is gated through AND gate 89a to the input 94a of strobe lamp 93a. When flip flop 85 is in its other state, a pulse is sent along wire 53b to input 94b of strobe lamp 93b. Thus, lamps 93a and 93b are alternately illumi¬ nated.

The field of view from the strobe lamp 93a with right hand circular polarizer 30a is shown as 22a. The illumination area from strobe lamp 93b with left handed circular polarizer 30b is shown as 22b. The video camera 80 has a field of view 23 that covers the overlapping region between 22a and 22b. In the video camera field of view 23 are the wing 15 with iced area 16. The images that corre- spond to wing 15 and iced area 16 that are shown on the video monitor 72 are labeled correspondingly as 15a and 16a.

During operation, the pulse generator 75 is set to provide trigger signals at a constant rate, e.g., in a range between 1 and 10 Hz. When a trigger signal enters the frame grabber input 77, it is synchronized with the frame grabber internal cycle and at the proper time the frame grabber pro¬ vides a strobe to flip flop 85 which is passed on to strobe lamps 93a or 93b. The strobe output is timed to be properly aligned with the frame synchronization signal that is sent along cable 84 from frame grabber output 83 into the video camera 80. Cable 84 provides a path from the frame grabber 70 to the video camera 80 for synchronization and a return path from video camera output 82 to frame grabber for the video signal.

If the pulse received by the AND gate 89a is enabled because flip flop output 85 is high, the strobe will pass through AND gate 89a, enter the strobe input 94a and

fire the strobe lamp 93a. The strobe lamp will produce a very short light pulse of approximately 10 microseconds length. The light pulse from the strobe lamp 93a illumi¬ nates the wing area. The reflected light from ice free specular area of the wing will be left hand circularly polarized because of the right hand circular polarizer 30a at the output of strobe lamp 93a. Because the video camera 80 has a right hand circular polarizer 40 at its input, it acts as part of an isolator. That is, any reflection from a clean metal specular area of the wing will reflect left hand polarized light which will not be able to get through the right hand circular polarizer 40 of the camera 80 and thus these areas as viewed by the camera will be very dark. The image sent by the video camera to the frame grabber will also appear very dark as well as the stored image that is sent from the frame grabber buffer memory into the video monitor 72 input 79 via wire 71.

Where there is ice present on the wing it will spoil the circular polarization of the polarized incident light and the image scene of the reflective light picked up by camera 80 and viewed on monitor 72 will not be dimmed. When the strobe signal passes through AND gate 89a, it simultaneously resets flip flop 85 to the opposite state such that AND gate 89b is enabled. Therefore, the next pulse from the pulse generator 75 into the frame grab¬ ber 70 causes the corresponding strobe pulse to be generated which will be gated through AND gate 89b to energize strobe lamp 93b whose light output is left hand polarized. Energy from strobe lamp 93b that strikes the wing 15 and returns from clean metal will be sent into the right hand polarizer

40 of the video camera 80. However, in this case, because the polarizations are of opposite hand, the reflected light energy that enters from specular reflecting portions of the wing 15 will pass through right hand polarizer 40 and into video camera 80 via lens 81 with only minor attenuation. That is, light from the left handed circularly polarized source 93b, 30b is changed to right handed circular polar¬ ization upon specular reflection from wing 15 and this light

may pass freely through the video camera's right hand circu¬ lar polarizer 40.

The corresponding analog signal from the video camera that is sent to the frame grabber 70 is recorded in its frame memory buffer and is output along line 71 to the video monitor 72. This particular signal will create an image that has little difference in light intensity between a specular area or an ice covered area. Polarization in this case is not important since the specularly reflected left handed circularly polarized light will pass through the video camera's right handed polarized filter 40. Thus, specular reflected returns and also the returns that come from paint or ice covered surfaces will pass equally well. Accordingly, the blinking effect will be produced for the area of a metallic surface which does not have ice on it. Video camera 80 is preferably of the type with a built in electronic shutter such as the Hitachi KP-M1. Because the camera shutter can be set for a very brief time interval that corresponds to the time interval of the strobe lamp illumination, the camera will be especially sensitive to the bright light from the strobe lamps and very insensi¬ tive to background light which will not be at a peak during the brief open shutter interval and will ignore all back¬ ground light outside of the interval that the shutter is open.

Figure 4B shows an indirectly viewed video based ice detecting system employing one strobed laser spotlight that makes it suitable for use with high background illumi¬ nation levels. In Figure 4B a strobe lamp 103a is a pulsed laser which typically has an output at a wavelength in the region of about 800 nanometers. Light from laser strobe lamp 103a is sent through a right hand circular polarizer 30 and covers the field of view 22a. The light from a laser is often naturally linearly polarized without using a linear polarizer and in such a case it may be circularly polarized by incorporating a properly oriented quarter wave retarder plate in right hand circular polarizer 30. The right hand polarizer 30 must be rotated to the proper position so that

its self-contained linear polarizer is in line with the polarization of the laser lamp output in the case that the laser light is naturally linearly polarized.

Video camera 80 views the scene via a narrow band interference filter 104 which is centered about the laser 103a output wavelength. Generally, such a filter will have a bandpass of approximately 10 nanometers and reject all light outside of the bandpass wavelength.

Reflected polarized light from the specular re- fleeting part of wing 15 entering the video camera 80 also passes through a rotating right hand circular polarizer 140 placed in front of the video camera lens 81. The rotating right hand polarizer 140 is driven by a motor 151. A signal from an encoder 153 attached to the motor 151 is sent only when the rotating right hand circular polarizer 140 has its plane parallel to the lenses 81 at the video camera input so that the optical axes of such lenses and of the polarizer are in alignment.

The analog video signal from the video camera is sent to frame grabber 70 input 83 via cable 84 and on the same cable the frame grabber synchronizing outputs are sent to the video camera input 82. A monitor plus VCR (optional) 72 is connected to the frame grabber video output via cable 71. The image of the wing 15a and the image of ice area 16a on monitor 72 corresponds to wing 15 and ice area 16 which are in the field of view of both the illumination patterns 22a from laser strobe lamp 103a and camera field of view 23.

In operation of polarizer 140, synchronous motor 151 rotates the right hand circular polarizer 140 in front of the video camera 80 at a high rate of approximately 600 RPM. The plane of the right handed circular polarizer 40 lines up with the lens plane of the video camera lens 81 twice per revolution. Thus, there are 1200 times per minute that a picture may be taken. The output from an encoder on the rotating shaft of polarizer 140 is used to identify each time that the rotating polarizer passes through such an aligned position. The two positions per rotation are alter-

nately isolating and non-isolating and correspond to the Figure 1A and Figure IB illustrations of the isolating and non-isolating modes achieved by turning one of the circular polarizers. The synchronizing pulses from the shaft encoder are sent to a programmable binary counter 160 which can be set to divide by any desired number. The output pulses from the binary counter are sent to the trigger input 77 of the frame grabber 70 along wire 73. A typical counter divides by any integer between 1 and 16. Counter 160 allows the rate at which pictures are taken to be adjusted from a very rapid rate to a slow rate. For example, the rate at which pictures will be taken when the divider is set to 16 will be 1200 pictures per minute divided by 16. To insure that alternating isolating and non-isolating states are obtained it is necessary to use only odd numbers as the divisor.

In both cases of Figures 4A and 4B electronic cir¬ cuits are preferably used to gate the video camera 80 to accept light only during the active interval of the strobe light. Also, in both cases optical bandpass filters may be used in front of the camera to match the strobe lamp's peak wavelength while simultaneously blocking out most of the wavelengths associated with ambient lighting. The typical strobe light, a xenon flash tube, produces a 10 microsecond flash which may be synchronized to the 1/10,000 second shutter of a commercial CCD video camera. Since the unshuttered camera would normally integrate ambient light for at least one field, or 16 milliseconds, there is an improvement factor of 160:1 - the effect of ambient sunlight can be reduced by 160:1. This 160:1 factor can be further improved by matching the strobe lamp (or pulse laser source) with a filter that cuts down the ambient wide band light by a much greater amount than the illumination source.

In both the systems of Figures 4A and 4B the video from the video camera is captured in a frame grabber and displayed on a video monitor. Thus, if the system alter¬ nates between the isolator and non-isolator state at a 2.22Hz rate (division = 9) the picture on the monitor will

be updated every 0.45 second and the human observer watching the monitor will see the ice free metallic surfaces blink between dark and bright at the 2.22Hz rate.

The embodiment of Figures 4A and 4B effectively add an image processing computer which performs arithmetic operations on individual pixels in multiple frame stores; one frame store per captured picture. The ability to per¬ form operations on pixels allows working with portions of the image that are of low intensity and also provides fur- ther means for eliminating the deleterious effects of unde¬ sirable background illumination such as sunlight.

Even if a curved aircraft surface region is illu¬ minated by multiple illumination sources of circularly polarized light, it will be found that due to the varied orientations of the surface normal with respect to the illumination sources and receiver there will be bright regions and dim regions in the image of the aircraft sur¬ face. The bright regions will correspond to those areas where the surface normal has the proper orientation to directly reflect the light from at least one of the illumi¬ nation sources into the camera lens. The dim regions corre¬ spond to those areas of the aircraft surface where the surface normal is such that the light from the illuminators is reflected predominantly away from the camera lens. As previously described, portions of the image that correspond to an ice free surface and are brightly lit will tend to vary between white and black in successive pictures on the monitors of the Figure 4A and 4B apparatus. However, por¬ tions of the aircraft surface that are ice free but in a dim region will vary between very dark gray and black in succes¬ sive pictures and so may be difficult to identify. This problem will exist both because of the limited dynamic range of the monitor and camera and because the ratio of dark to light is intrinsically less for off axis returns. Any remaining background due to sunlight further reduces the apparent brightness ratio between ice free regions of suc¬ cessive images, particularly in the dim regions, by adding

unwanted illumination to the images taken in both the iso¬ lating and non-isolating mode.

An optimum use of the equipment of Figure 4A and 4B is to first capture an image in the frame store that corresponds to strobing the illuminator but blocking the light from specular reflection from ice free metal; e.g., capture a picture in the isolator mode. Next, the illumina¬ tor is strobed and a picture is captured in the non-isolat¬ ing mode. Finally, a picture is captured with the illumina- tor strobe off; this captures a picture that consists purely of the undesirable background light. If the receiver (de¬ tector) optics is being varied between pictures to change between the isolating and non-isolating mode of operation, it is not important which mode it is in when the background image is captured because both modes will have been balanced for equal light attenuation of unpolarized light.

The digital value corresponding to the background illumination in each pixel of the frame grabber holding the background may now be subtracted from each corresponding value in each of the pixels of the image in which specular returns were blocked; e.g., from each pixel of the image taken in the- isolating mode. The process of subtracting the background is repeated for each pixel in the frame grabber holding the image taken in the non-isolating mode. At this point, assuming linearity of the pixel values, the effect of any remaining background light has been removed from the two frame stores. If the recording or digitizing process is not completely linear the non-linearity must be removed before performing the subtraction. This is normally performed at the time the image is first digitized and entered into the frame store via the use of a look up table in the image processor and is well known in the state of the art.

Once the images in the frame grabbers have had the effects of background illumination removed, the image pro- cessor can find the ratio of amplitudes between correspond¬ ing pixels in the two images. By forming a ratio of the value of the intensity of the pixels in the second (non- isolating) divided by the value of the intensity of the

corresponding pixels in the first (isolating) a ratio having values generally equal to one or greater than one will be obtained. Ice free metallic surfaces that have surface normals reflecting the illumination towards the camera lens will have the highest ratios. A normalizing value approxi¬ mately equal to the Nth root of one divided by the largest of the two pixel values that created the ratio (generally, the value of the pixel from the non-isolating picture) may be used as a multi-plier to enhance the ratio from the ice free surfaces that are dim due to their being off-axis with respect to directing the reflected light towards the camera. N is typically an integer equal to or greater than 2. Of course, only values higher than some chosen threshold should be so normalized so that the system does not respond to noisy signals. If desired, the preceding arithmetic manipu¬ lation of pixel values may instead be performed on groups of pixels that correspond to segmented and/or filtered portions of the aircraft surface image. These filtering techniques which include low pass spatial filtering and median filter- ing may be used to operate on noisy images and are well known in the state of the art. Another suitable metric for comparing corresponding isolating and non-isolating pixel or region brightness amplitudes is the normalized difference. This may be formed by subtracting corresponding pixel or region amplitudes and dividing the result by the sum of their amplitudes.

To highlight ice free regions in the most easily interpreted form, the ratios may be assigned to colors as, for example, that high ratios are assigned to the color green, low ratios to the color red, and intermediate regions with the color yellow. These colors may be used to color the non-isolator image on the screen of the color video monitor. Optionally, the ratios may be encoded in black to white intensity levels that may be displayed in the same manner as the color encoded images. Such levels may be used to indicate ice thickness according to the amount of depo¬ larization observed.

All of the preceding techniques of using isolator and non-isolator structures may be implemented by using linearly polarized light in the illuminator, rather than circular, and equipping the receiver (detector) with a linear polarizer that is alternately aligned with and then at right angles to the polarizer in the illuminator. This mode of operation depends upon the fact that an ice-free metallic surface will return polarized light approximately unchanged whereas an ice covered metal surface or matte material will de-polarize the light. Thus, once again, an ice covered metallic surface will remain at approxi-mately the same intensity. Of course, the transmitted linear polarization can be alternated between being aligned with and then being at right angles to the direction of a linear polarizer in the receiver to achieve the same end.

Figure 5A shows the details of the Figure 4B rotating circular polarizer 40 and video camera 80 assembly. Video camera 80 is mounted to a bracket 150. A motor 151 is also mounted to bracket 150 and has a slotted output shaft 152 for holding the circular polarizer 40 to rotate in synchronism with the shaft. An encoder disk 153 mounted on shaft 15 is used to sense the position of the rotating polarizer 140. Encoder disk 153 has a photo optical inter¬ rupter 154 supported by a member 157 affixed to bracket 150. The encoder disk is solid everywhere except for two posi¬ tions, 180° opposite, which are in line with photo inter¬ rupter 154 only when the optical plane of polarizer 140 is parallel to that of the lenses in video camera lens assembly 81. A top view of this arrangement is shown in Figure

5B and Figure 5C which shows an encoder pickup 154 which incorporates an LED light source and a photo diode in one package that is commercially available as Optek part number OPB120A6. - Figure 6 is an optical schematic of the laser diode spotlight assembly. Light from laser diode 300 is collected by a collimating lens 301 and the collimated beam is sent into the telescope formed by a negative lens 302 and

a positive lens 303. When the focal point of the positive lens is coincident with that of the negative lens, a colli- ated beam emerges from the positive lens 303. Positive lens 303 is shown in position 310 so that its focal point coincides at 312 with that of negative lens 302. A colli- mated beam 304 is the result of this configuration. When the positive lens 303 is moved closer to the negative lens, such as to position 311 in Figure 6, the beam 306 that emerges, is expanding and so can cover a wider field of view. Thus, by adjusting the position of the lens from 310 to a point where it close to the negative lens, it is possi¬ ble to obtain any output light cone between collimation and a cone slightly narrower in angle than that of the beam 313 as it leaves negative lens 302. The arrangement of Fig. 6 is also applicable to all other illumination sources shown when the source (filament or flash lamp) is small.

Figure 7 shows an indirect viewing system for ice on metal detection that uses a synchronous detection method that can operate with one photosensor or an array of photosensors, according to the field of view and resolution required. The illumination source for the surface area 15 to be inspected is not shown but it may be any bright source of either right handed or left handed circularly polarized light. The area that is to be inspected is imaged via camera lens 400 onto photodiode 402, or onto an array of similar photo diodes. The circular polarizer required for isolation is formed by quarter wave retarder plate 42, linear polarizer 41 and Verdet rotator 401. The Verdet rotator is typically of garnet and energized by a magnetic field created by a coil 450 via power buffer amplifiers 402a and 402b which alternately drives current through the sole¬ noid, first in one direction and then in the other. The effect is to cause linearly polarized light from the polar- izer passing through the rotator to change the direction of its polarization by plus or minus 45° according to the direction of the solenoid current flow. Other devices based

on the Hall and/or Pockel's effect which use high voltage fields could be used in a similar manner.

In the static condition with no current flow through the coil 450, both right handed and left handed reflected circularly polarized light from area 15 will pass through the rotator 401 to the photo diodes 402 with little attenuation because the slow axis of quarter wave plate 42 is in line with the polarization axis of linear polarizer 41. Therefore, light of either hand circular polarization is at 45° to the polarizer and so can pass through rotator 401 without large attenuation. However, when the coil is alternately energized with current flow in opposite direc¬ tions, the addition and subtraction of 45° to the plane of polarization present at the output of the quarter wave plate 42 causes the plane of polarization to alternate between vertical and horizontal at linear polarizer 41. Thus, reflected circularly polarized input light will alternately be allowed to pass and not pass to the photo diode detector.

Because the rotation of the plane of polarization is performed via current direction switching, it can be performed quite rapidly. A 10 KHz rate, which is adequate for the apparatus, is easily obtained. A clock source 406 provides pulses to a flip flop 403 at its toggle input 405. The flip flop 403 outputs 404a and 404b are amplified by buffers 402a and 402b to energize coil 450 in a direction that varies according to the state of the flip flop 402. The optical energy received at the photo diode array 402 generates a corresponding electrical signal that is applied over input line 408 to a differential amplifier 407. The output 409 of amplifier 407 feeds two buffer amplifiers 410a and 410b via their inputs 41la and 411b. Both amplifiers 410 have equal gain but are of opposite polarity. A multiplexer 413 has its inputs 414a and 414b connected to the two amplifier 410a and 410b outputs 412a and 412b. The multiplexer 413 directs its two inputs to its single output 415 according to the state of its select

terminal 460 which is connected to output 404a of the flip flop 403. The output 415 of the multiplexer 413 is applied to an integrator 416 or optional low pass filter 426. The integrator 416 is formed by input resistor 420, operational amplifier 417, capacitor 421 and field effect transistor 422 which is used to periodically reset the integrator by dis¬ charging the capacitor. This arrangement is well known in the art. The integrator 416 (or filter 426) output 419, when greater than a threshold voltage positive or negative as set by a double end zener diode 480 will energize one of the oppositely poled LED's 424a or 424b.

The detection circuit of Fig. 7 rejects the light reflected from diffuse or ice covered areas but passes that from ice-free specular surfaces. Diffuse or ice covered surfaces return unpolarized light to the detector. With these type surfaces, although the current direction in the Verdet rotator 401 is changing direction at a 10 KHz rate, the light received by the photo diode 402 remains at a constant level - the light amplitude is unchanged because the light is not polarized.

The electrical voltage at the output 409 of ampli¬ fier 407 responds to the input level and remains constant. The multiplexer 413 alternately selects equal constant level positive and negative voltages so that the integrator 416 (or low pass filter 426) output stays close to zero and neither of the LED's 424a or 424b draw current since the output voltage does not overcome the zener diode 480 thresh¬ old voltage.

It can be seen that when area 15 is ice free the light returned to the apparatus will be circularly polarized and the signal at photo diode 402 will alternate between a large and small value at a 10 KHz rate. Since the two voltages selected at terminals 414a and 414b will differ in amplitude, they will not average to zero at the output of the integrator 416 (or low pass filter 426) and one of the LED's will light, according to whether the larger of the two voltages at point 409 was received when the state of flip flop 403's output 404a was high or low. This, in turn

usually depends upon whether a right handed or left handed illuminator is being used. The output LED can also change if the area being observed 15 receives most of its circular¬ ly polarized illumination indirectly via specular reflection from another surface, since each such reflection changes the state (hand) of the circular polarization.

The apparatus of Figure 7, when used with a single photo-detector is useful with a mechanical drive apparatus that scans the optical axis of the assembly in both eleva- tion and azimuth to generate a raster scan which will create a full image of a scene on a point by point basis. The output 419 may be sent to a video display which is being scanned via its deflection circuits in synchronism with the mechanical drive apparatus to paint the image on the screen. As an alternative, the optical axis may be scanned in a raster pattern using azimuth and elevation deflecting galva¬ nometer arrangements such as are available from General Scanning Corporation. Of course, such synthetically gener¬ ated images may also be digitized and processed using the image processing hardware and software techniques previously described.

Figure 8A shows an embodiment of the invention useful when it is important to obtain ice detection informa¬ tion in an extremely rapid mode that is useful for scanning across an object in a short time without the smearing or the misregistration that may occur when the camera is panning and sequential pictures are taken for the isolating and non- isolating modes.

In Figure 8A, a strobe lamp 103a is used with linear polarizer 501 to illuminate the surface 15 via polar¬ ized light cone 22a. The video camera 80 lens 81 images the scene as contained in field of view 23 which overlaps cone 22a. A polarization preserving beam splitter 503 is used to divide the energy received by lens 81 into two substantially equal amounts which are directed to video cameras 80 and

80a. Camera 80 is fitted with linear polarizer 500 which is in alignment with linear polarizer 501 so that reflected specular energy may pass with little loss and so creates a

non-isolating mode receiver. Camera 80a also is fitted with a linear polarizer 501 but its axis is aligned at 90 degrees to that of linear polarizer 500 so that reflected specular energy is blocked which creates an isolating mode receiver.

When the synchronizing pulse is received via wire 87, the strobe lamp 103a flashes for a brief time; 10 micro¬ seconds is typical. During the brief flash interval the isolating and non-isolating images are captured on the silicon CCD devices (typical) in the two cameras, 80a and 80, respectively. The two images can be read out sequen¬ tially via a multiplexer and recorded in the digital frame buffers of the image processor. A multiplexer of the type required is built into the Cognex 4400 and is normally part of most commercial frame grabbers and image processors. The processing of the images is substantially the same as previ¬ ously described with amplitude comparisons being made be¬ tween corresponding pixels or corresponding regions.

Because the two cameras use a common lens 81 the images will have top and bottom reversed (one is viewed through a mirror) but are otherwise substantially geometri¬ cally identical. Calibration may be obtained by recording any two points in the field of view and mechanically adjust¬ ing the CCD chips via translation and rotation to have a one to one correspondence of pixels. This can also be accom¬ plished via software within the image processor and such conventional software is normally furnished with the image processor. Because the lens 81 and cameras 80 and 80a are held in alignment, the calibration, whether via mechanical or software means, need only be performed once, at the factory.

In Figure 8A, the linear polarizers may be re¬ placed with circular polarizers such that at least one of the circular polarizers in the receiver has the same "hand" as- that of the transmitter to provide an isolating mode image and the other has the opposite "hand" or not be circu¬ larly polarizing and have suitable attenuation to ensure that diffuse objects have the same intensity in both pic-

tures. Additionally, if polarizing beam splitters are used, one or more of the polarizers in the receivers may be omit¬ ted since polarizing beam splitters will divide energy according to polarization properties. In Figure 8A, the isolating and non-isolating images may be obtained with two separate cameras as shown, but with two separate and substantially matched (in focal length and axis parallelism) lens means, one per camera, that create geometrically corresponding images. The corre- spondence need not be exact if corresponding image features or regions or pixel groups are compared with respect to average amplitude in the isolating and non-isolating mode.

An alternative arrangement shown in Figure 8B, section view, uses a mirror 510 in each path and so does not invert one image with respect to the other.

As can be appreciated, the camera in all embodi¬ ments may be replaced with a multiplicity of cameras at various positions and angles to the illuminated surface to gather more of the specularly reflected light and similarly, a multiplicity of illuminators may be used at various posi¬ tions and angles to the illuminated surface to assist the cameras in gathering more of the specularly reflected light. It is only necessary that when such arrangements are used that all control signals and polarizers be common to the group of cameras that replaces one camera or to the group of illuminators that replaces one illuminator.

The arrangements of Fig. 8A and Fig. 8B require multiple cameras and beam splitters which are similar to first generation color cameras which employed three separate cameras to separately record three separate images, one for each of the primary colors. More modern color cameras employ a single camera with a patterned color filter that is organized in closely spaced columns; e.g., R,G,B,R,G,B,R,G,B.... where R represents red, G represents green and B represents blue. This has the advantage of using only one camera plus simple electronics and requires a one time adjustment of the filter to the camera chip at the factory. The same identical color camera pickup chip and

electronics circuits may be used to manufacture a polariza¬ tion sensitive camera by replacing the tri-color filter used in the color camera with the two layer filter shown in the assembly of Fig. 9A. In Fig. 9A, the camera pickup is represented by

CCD chip 900 with typical scan lines 901. A thin linear polarizer 910 with polarization axis at 45 degrees to the "slow" axis defined for patterned retarder plate 920 is located touching, or in close proximity to the illuminated surface of the CCD chip. Retarder plate 920 is manufactured from a birefringent material and selectively etched so that adjacent columns differ by 1/4 wave with respect to the retardation produced and a pattern of +,0,-,+,0,-,+,0,-,... is maintained where + represents +1/4 wave (923) , 0 repre- sents equality of phase (922) , and - represents -1/4 wave (921) . The patterned retarder plate must be in close prox¬ imity to the polarizing plate. The retarder plate selective etching may be done chemically or with ion beams and is well known in the semiconductor industry. The process is cur- rently being used to create micro lens arrays known as binary optics.

The arrangement shown in Fig. 9B requires two mask and etch steps to obtain the three thicknesses needed for manufacturing the three retardations needed for the three column types. The optional filling 924 adds non birefrin¬ gent material having an optical index approximately equal to that of the birefringent material to provide the overall structure of a thin glass plate with respect to a focused light beam. As shown in Fig. 9B, the columns are brought into alignment with the pixels 905 in a CCD column in exact¬ ly the same manner as is done for a color camera.

In operation, the polarization images produced by the patterned retarder plate will be processed by the color camera's electronic circuits into either three separate images or a single composite image. In the case that a single composite image results, it can be decoded by any color receiver into corresponding R,G,B images which will represent not the three colors but the three states of

circular polarization received which correspond to left, right and non polarized. These images may be processed according to all of the preceding methods regarding ice detection. Although all cameras shown have been of rectangu¬ lar format, in some circumstances it may be preferable that a linear camera array (single row of pixels) be used and the field of view be transversely scanned via rotating polygon mirrors, galvanometers, rotating prisms, or other scanning means to synthesize a rectangular image of some desired format. At such times the illuminator may provide a "line of light" which would be likewise scanned in synchronism with the scanning of the linear array. This is suitable for fields of view which may be long and narrow and require more resolution than may be obtained from the standard camera format.

Figure 13 shows a scanning ice detection system. A laser diode light source 13, such as Spectra Diode Labora¬ tories SDL 5422-HI capable of producing short, bright pulses of light is projected onto path folding mirror 18a which reflects the light pulses through polarizing prism 603 (a linear polarizing plate could be used when using just one light source 13) . Prism 603 linearly polarizes the light, reflecting the unwanted orthogonal polarized light to be absorbed by absorber 17. The linearly polarized light passes through quarter wave retarder plate 12a, converting the light to circular polarization. Mirror 18c mounted on lens 81 folds the light path to coincide with the receiver light path. A galvanometric scanner 217, such as Laser Scan¬ ning Products GRS-PS, series scans the light over an angle 219 in the horizontal plane via oscillating mirror 218. Scanner 217 is rotated in the vertical plane by motor 63 driving shaft 64, causing the horizontally scanned light to also scan vertically. Light reflected by a surface illumi¬ nated by the scanned light retraces the path to a positive lens 81 which focuses the light in combination with a nega¬ tive lens 281 onto pinhole 282 in barrier 283. Light pass-

ing through a pinhole 282 is focused by a lens 284 upon avalanche photodiodes 180 (such as the 5mm avalanche photo- diodes found in the Advanced Photonix APM-10 Detector mod¬ ules) after passing through quarter wave retarder plate 12b and polarizing prism 503. Mirror 18b folds the light path. Plate 12b converts the circularly polarized light to linear¬ ly polarized light, which if aligned with the polarization axis of prism 503 passes through to mirror 18b and one diode 180. If the polarized light is orthogonal to the polar- ization axis of prism 503, it will be reflected to the other diode 180 by prism 503. Unpolarized light will illuminate the photodiodes 180 equally whereas proper alignment of the circularly polarized light from light source 13, prism 503 and plate 12b can produce a maximum difference in light levels on one diode 180 relative to the other diode 180 when light projected by the system is specularly reflected. Narrow band interference filter 104, centered about the wavelength of light source 13 reduces the amount of ambient light which reaches photodiodes 180. Quarter wave retarder plates 12a and 12b may be removed from the system and the system will then use linear¬ ly polarized light to produce a large ratio difference in light levels on photodiodes 180 for specular reflections and equal light levels on photodiodes 180 for unpolarized re- flections.

An alternative system adds a second light source 13 which passes light aligned to the polarization axis of prism 603 onto absorber 17 and projects light orthogonal to the polarization axis of prism 603 which is reflected by prism 603 through plate 12a along the same path as the other light source 13 by careful alignment. The polarizations of the light from the two sources 13 are of opposite hands so that only one diode 180 is required to detect the large ratio of light reaching the diode from the two sources when reflected by a specular surface. The unpolarized light reflected from either source 13 will reach a diode 180 with equal intensity if the source 13 levels are equal. The galvanometer's mirror will typically scan 3 meters in 1/400

sec or equivalently travel 0.012 cm in the time between strobing the two light sources 13 100 nsec apart. Thus the diode receives light reflected from essentially the same area from the two sources. As previously noted, plates 12a and 12b may be removed.

When one light source 13 and two photodiodes 180 are used, the diode receiving the large specular reflected light may tend to heat up and alter any calibration of signal level. Using two diodes and two light sources 13 can reduce this problem. By alternately strobing the light sources 13, the heating will be equal since the large specu¬ lar energy will alternate between photodiodes 180.

Although the equipment described separates clear wing from ice and snow, it does not separate (except visual- ly to the operator's eye) runway and other background sur¬ faces from wing surfaces, etc. This can be done via image processing techniques or stereo ranging or lidar (optical radar) ranging. Also, image processing techniques to be employed can segment surfaces of like texture and only color red those "non-blinking" areas that are substantially sur¬ rounded by "blinking" areas (green) . That is, ice would be highlighted only when substantially surrounded by clear metal. As an alternative, stereo ranging may be used to separate foreground from background and only the foreground (wing or other aircraft surface) have non-blinking areas tagged to highlight ice formation.

When viewing downward on an aircraft wing, the ground appears in the field of view and returns unpolarized light similar to ice. One method of rejecting this unwanted signal is to use the time of travel of the light pulses to determine which surface is reflecting the light. If the wing is at least 5 feet above the ground, the ground signal will reach the diode 180, at least 10 nsec later (curve 3, Figure 14) than a signal from the wing (curve 2) relative to the time of the strobe 289 (curve 1) . Thus any signal exhibiting this additional delay can be rejected as not belonging to the wing. Since the measurement can be made on the leading edge by measurement unit 286, pulses wider than

10 nsec can be used without effect on this rejection pro¬ cess.

The problem of large dynamic range of reflected light from specular surfaces is not as great for the scan- ning system since the dynamic range of avalanche photo diodes (such as used for the APM-10 detectors) is much larger than that which is available for most imaging camer¬ as. To deal with the large dynamic range of the output of these diodes, it is preferable to use a logarithmic amplifi- er 280 such as model AD640 from Analog Devices. Since ratios of signal levels are being analyzed, the log ampli¬ fier has the additional benefit of producing the same volt¬ age change for a fixed ratio throughout the dynamic range. A further problem, common to both scanning and non- scanning systems, is that the ratio of measured values in the isolator and non-isolator states reduces towards that of ice as the viewing angle deviates significantly from normal incidence when viewing a specular surface. This increases the difficulty in processing signals over a wide dynamic range.

To further overcome the influence of wide dynamic range on the display produced by the system, the display of the images can be enhanced by processing via computer 287 the signal level produced by each pixel in the image of camera 80 or by photodiodes 180 and to quantize the display for that pixel (or scan point for the scanning system) into three levels: clear, ice and non-ice. A background refer¬ ence level is obtained for the pixel (or scan point) when the light source 13 is not being strobed. This will account for any ambient light. The short strobe time possible with the scanning system essentially eliminates interference by ambient light. The measurements are then made for the pixel (or scan point) by strobing twice; once with the optical means in the light path between light source 13 and camera 80- (or diode(s) 180) in an optical isolator state and once in an optical non-isolator state. A predetermined table of threshold values as a function of the measured background reference level and the value measured in the non-isolator

state is stored in computer 287. A further problem that can be addressed by the table is the reduced ratio of measured values obtained in the isolator and non-isolator states when viewing a specular surface significantly away from the surface normal. If the ratio of the non-isolator state value to the value measured in the isolator state exceeds the threshold value from the table, the pixel is declared to be in a clear area, otherwise it is declared to be in an ice area. If the isolator state measured value in the ice area is less than a given value, the pixel is declared to be in a non-ice area. By displaying on display 72 the three catego¬ ries as black, white and grey, or various contrasting col¬ ors, the display readily conveys the desired information concerning the icing condition on wing surface 15. When multiple levels of illumination are used to increase the dynamic range of the system, the above method for quantizing the measurements into three levels will tend to declare "ice" for one or more of the levels when the declaration should be "clear". Thus if a declaration of "clear" is made at any level of illumination, the pixel is declared in a first decision to be in a clear area. The measured value in the isolator state with the highest illu¬ mination is used in a second decision to determine "non-ice" in areas not declared "clear" by the first decision. If time of arrival is used to eliminate ground returns, then those signals of a continuous group that are received sig¬ nificantly later than signals of another continuous group are declared "non-ice". A process flow diagram is provided in Figure 16 for the steps described above. Figure 13 provides the details of the scanning system signal flows indicated above. Timing and control unit 285 synchronizes strobes 289 to the oscillating mirror 218 (synchronizing signal not shown) . Strobes 289 cause light sources 13 to emit light pulses. Logarithmic amplifi- ers 280 receive diode 180 output signals 293 and send their compressed dynamic range outputs 290 to analog to digital converter 288 and leading edge measurement unit 286. Lead¬ ing edge measurement unit 286 starts measuring time at the

time of strobe 289 and stops when input signal 290 exceeds a given value. The time interval thus measured is transmitted to computer 287 via signal 294. Analog to digital converter 288 reports the amplitude measured on input signal 290 to computer 287 via signal 291. Computer 287 quantizes input signal values 291 into categories of clear, ice and non-ice as described above, using leading edge time on signal 294 to force background signals to be classified as non-ice. The categories are transmitted to display 72 via signal 292 from computer 287 to provide a visual image of wing 15 and ice patches 16.

Leading edge measurement unit 286 can use any standard processing means that provides adequate time reso¬ lution. Figure 17a illustrates the waveforms of an analog embodiment. Strobe 289 from Timing and Control unit 285 is shown as pulse waveform 171 which establishes time = 0 for measuring time of travel for light to reach surface 15 and return to photodiodes 180. A sweep voltage having waveform 172 is started at the leading edge of pulse 171. The waveform 173 of the output of amplifier 280 is compared to a threshold and triggers a track and hold circuit to sample the sweep voltage (or alternatively just stops the sweep) providing waveform 174. Waveform 174 is then converted to a digital value at time A by an A/D converter, where time A exceeds the maximum expected signal delay time.

Figure 17b illustrates an embodiment using a digital counter. Again, strobe 289 is shown as waveform 171 to establish time = 0. Waveform 175 indicates the waveform of gated clock pulses that are turned on by the leading edge of pulse 171 and turned off after the maximum expected signal delay time. When amplifier 280 output 290 shown as waveform 173 exceeds a threshold, a counter counting the clock pulses is stopped and holds its count as indicated by waveform 176 (alternatively the clock pulses can be stopped when waveform 173 exceeds a threshold) . At time A the counter value is sampled for use by computer 287.

The computer 287 can further improve the display by comparing the declared category of a pixel over several

scans. If an indication of "clear" and "ice" alternates, then it can be concluded that ice has not formed and the sporadic declarations of "ice" are caused by blowing snow which can be displayed as "clear" or a fourth category. As indicated above, a problem exists when viewing a wing surface at an angle significantly away from the surface normal. The ratio of the values measured in the isolator and non-isolator states approach that of ice, thus preventing reliable discrimination. Certain paints, materi- als and surface treatments can be applied to remove this deficiency.

Currently some aircraft wings are painted with a gray paint that provides a low ratio of isolator to non- isolator response. Adding 10% by volume of metal chips to the paint significantly improves the response but not as much as is desired. However, it has been found that a significant response improvement can be obtained by using black paint to which 10% by volume of metallic chips have been added. Surprisingly, an observer sees the appearance as similar to the gray paint, probably because the specular reflection of the metallic chips produces a gray appearance. Figure 15 shows the ratio of response obtained by the ice detection system when viewing various surfaces as a function of the angle the system makes to the surface nor- mal. The bare metal curve 1 starts at a ratio of 66 on the surface normal and drops to 3 at 12 degrees from normal. Measurements of paper (curve 3) which is similar to that received from ice starts at 1.55 on the surface normal and drops to 1.47 at 12 degrees from normal. Thus at least a 2:1 difference in ratios exists out to 12 degrees from normal, and reliable discrimination is possible. Beyond that, performance becomes gradually less reliable.

Curve 2 is for a paint with approximately 10% metal chips added by volume which, when applied to the surface, provides a ratio of 3 out to 30 degrees from nor¬ mal. Other currently existing aircraft paints provide a similar response.

Retroreflector tapes and paints that contain reflecting sites show very little sensitivity to the angle of view. Curve 4 shows that a surface covered by 3M silver provides a ratio greater than 27 out to 50 degrees from normal. 3M Reflective Highway Marking Tape series 380 (white) and series 381 (yellow) are used on roadways as reflectors and thereby provide the basis for detection of ice formation on roadways and bridges. Many of the suitable retroreflective tapes are manufactured by embedding tiny metallic coated dielectric beads in a clear plastic carrier. The tiny spherical beads reflect a portion of the impinging light back towards the light source, substantially indepen¬ dent of its direction. In fact, the reflected energy is generally much larger than required and can saturate the receiver. It is therefore preferable to add an attenuating layer of material to the surface of the retroreflective tape to reduce its response. An attenuation of approximately 3:1 each way is recommended as a best value (approximately 10:1 roundtrip loss) . Another method of reducing the sensitivity of re¬ sponse to angle from surface normal from a metallic surface is to cause the surface to present many small facets at all angles so that at any viewing angle a significant number of facets will have their surface normal essentially aligned to the viewing angle. Sand blasting, roll dimpling, etching and other methods are in common use to produce surfaces with this characteristic.

The invention is to facilitate ease of use and promote record keeping which is of vital importance to the aviation safety industry. Although not shown in the various drawings, it is anticipated that flight number, aircraft identification, time and date and other pertinent informa¬ tion would be aurally, visually, or textually annotated to the display monitors and to the disk or tape recordings made with the ice detection equipment. The performance of this task would be implemented with commercially available compo¬ nents that are often part of the equipment specified (earner-

as and recorders) or via additional "plug compatible" anno¬ tation and editing devices.

It may be desirable to locate the recording and viewing or control equipments at remote locations such as the aircraft cabin, control tower, ground control area, or aircraft terminal. Cameras and illuminators may also be built into various remote portions of the aircraft from which the wing or other surface is to be monitored. Accord¬ ingly, the various wires shown in the drawings, whether for purposes of data or signal transfer or control, may be replaced with telemetry equipment operating via radio, infra-red, power lines or fiber optic links.

In all claims and in the foregoing disclosure the term "light" is to be interpreted as "electromagnetic ener- gy" and not restricted to just the visible light portion of the electromagnetic spectrum inasmuch as the principles described are not so limited and in fact extend into the infrared and beyond.