The present invention relates to a scanning device, and particularly a scanning device for scanning the pages of a book, magazine, or other bound document into a suitable format to allow the document to be read.

Existing technology for scanning or copying pages from books are not well-suited to conveniently, quickly and accurately capturing multiple images. Existing well-known scanning/image capturing devices include conventional photocopier/Xerox® machines or flatbed scanners that are suitable for use in the office or home, but they are not suited for use in reading a book or magazine. Such existing devices are heavy and difficult to move and the reading process is relatively slow. There is a significant need for scanning devices for use by the visually-impaired or other users who have difficulty reading.

An example of a prior art device for use by the visually-impaired is that provided by Humanware Inc's "scan and read" products, incorporating automated reading technology. Such known devices use an overhead static camera and a lens to directly view and image a page. Each page needs to be laid below the camera at a specified distance on a flat surface; or a mechanical means is provided to scan over a page laid down in full contact with a plate or glass window. However, it has been found that the scanning/reading process using such devices is very slow and cumbersome, not least because of the key requirement of known devices that any pages are presented as a flat surface to the image reading element of the device. If the page is not presented as a flat surface, the image will be distorted and it is unlikely that the device will be able to read the page.

Portable readers include the "Intel® Reader", a handheld "point and shoot" device, which converts visual data into digital data and then reads the content aloud to a user. A similar portable device is available from the Scanning Pen Shop, which is a reading aid requiring a user to scan along the lines of text in a book. However, capturing the image using this device is a slow process and requires a visually impaired user to scan at the correct speed and be able to accurately aim at the data to be captured.

Known scanning devices, such as those disclosed in Japanese patent publications JPH07203156 and JPH05207245 require motors and/or mechanical drive systems to move linear image capture apparatus across the entire page to be scanned. Such motorised or mechanical drive systems limit the speed and constrain the size of known scanning devices, whilst also rendering them more prone to failure.

Existing scanning solutions may be useable for a single page document or a newspaper or magazine that can easily be folded or taken apart. However, if a user wishes to read a thick book with a stitched spine, or a delicate book then there is a real likelihood that either the book will be damaged or the required pages cannot be imaged. This is particularly difficult for pages in the middle of a book or when a book is new because it is difficult to open the book to the required extent without stressing the spine. Furthermore, if books are not opened fully then illumination variation across the page of the book makes it highly likely that any captured image will be distorted. A distorted image would not be suitable for processing into any further format; for example if the image is to be converted into audio output then the audio output will also be "unreadable" or inaccurate.

DE102009013831 discloses a wedge-shaped device for scanning bound documents comprising a linear sensor contained within an imaging optical/scan system. The use of a linear sensor means that it is necessary to move the entire optical system transversely across the entire surface of the page to be scanned, with the direction of motion of the scan system parallel to the support surface. DE102009013831 discloses that only one of the two open pages of the book are scanned and that the beam path is angled relative to the contact surface, which is less that the angle of opening of the device. JP20131060508 discloses an image reading apparatus for use in a copier, printer, scanner or facsimile. The device comprises two planes of glass separated by an angle of 90 degrees, which would require stress to be placed on the spine of the book for scanning. The device forms a triangular prism-shaped contact surface against which a paper document to be scanned must be placed (without the paper "floating" from the contact glass). JP2013060508 requires the entire assembly of a light source, mirrors, lens and linear image sensor to be moved across the full width of an open book, i.e. a drive system is required. The device is not portable and is too large to be hand-held.

US5359207 discloses a wedge-shaped scanner utilizing two dimensional sensing arrays to simultaneously scan both pages of an open book. The scanner offers an alternative solution using expensive two-dimensional detector arrays that require substantial contact with a page to be read. The detector arrays used in US5359207 would make the device unlikely to be portable and do not require an imaging assembly because each detector pixel only receives light from the directly adjacent page area.

The present invention sets out to provide an improved scanning device, which alleviates the problems described above to provide a much improved portable scanning device, particularly suited for use by the visually-impaired or those with reading-based difficulties.

In one aspect, the invention provides a scanning device for inserting into the pages of a bound document to read text contained therein; having at least two scanning surfaces separated by an angle of between about 15 and about 20 degrees to extract data from each of two opposing pages of the bound document; wherein the scanning device is substantially wedge- shaped and comprises a plurality of static imaging assemblies, wherein each imaging assembly comprises a sensor; a lens; and a moveable focussing housing ; and a processor to process each of multiple images, generated by the or each imaging assembly, to correct for the angled field of view and produce a composite full-page image. It is understood that in the context of the present invention "scanning" is to systematically examine or read the document. The present invention offers a significant advantage over the prior art because there is no requirement for a motorised or mechanical drive systems to move the entire imaging assembly over the entire page to be scanned. Rather the imaging assemblies themselves are static; i.e. the position of each imaging assembly with regard to the scanning surfaces is constant. Only the focussing housing of the lens is movable to allow for multiple images to be obtained at each of a plurality of different, pre-set positions of the lens with respect to the sensor. The movable focussing housing of the present invention is required to move a distance of less than about 50 micrometres to achieve the required focussing, preferably moving less than about 10 micrometres or even a sub-micrometre distance, contrasting with the requirement of known devices to move the entire scanning/imaging assembly across the full length or width of the page to be scanned. This allows the present invention to be more compact and more reliable than known scanning devices using motorised or mechanical drive systems. The present invention allows for text, graphics, captions etc. to be "read" by visually-impaired users or users with reading-based difficulties (such as dyslexia). Books, magazines and other bound documents can quickly, conveniently and accurately be scanned and the collected data can be translated into any required format; for example, the captured image can be magnified and projected; stored onto disc for later retrieval or presented as audio commentary to a user.

It is also understood that in the context of the present invention the "wedge-shaped" device has a solid shape defined by two triangular faces and three rectangular or trapezoid faces which form the scanning surfaces.

The angle between the scanning surfaces is well-suited to ease of insertion into a bound document even if the document is very thick or tightly bound. It is to be understood that "substantially" refers to a greater part i.e. the planar scanning surface does not deviate from a flat plane across more than about 10 % of its surface.

Within this specification, the term "about" means plus or minus 20%, more preferably plus or minus 10%, even more preferably plus or minus 5%, most preferably plus or minus 2%.

Preferably, the scanning device is substantially shaped to be a triangular prism.

By having a wedge-shape, the device can easily be held by a user and conveniently inserted into a bound document, such as a book, so that the scanning surfaces are in contact with or in close proximity to the page/s to be scanned. The wedge-shape of the device means that the device has a thicker end, which houses internal workings of the device, and tapers to a leading edge, which is a thin edge that can be inserted near to the book's spine. This ensures that the full page width can be scanned by the device. The shape of the device also gives a tactile indication to a user as to the correct orientation of the device for insertion into a book; i.e. if the user is visually-impaired. The device is easy to grip in the hand so that it can quickly and easily be moved through the pages of the book so that a user can enjoy the experience of "reading" the book in a manner comparable to reading the book with their eyes.

Preferably, each scanning surface is on a rectangular or trapezoid face of the wedge-shaped device.

Preferably, each of the two scanning surfaces is configured to scan one of each of two opposing pages of a bound document simultaneously.

Preferably, the or each scanning surface is substantially planar.

Preferably, the or each scanning surface is made from a transparent plastic. Preferably, the or each scanning surface is coated in an anti-reflective coating.

Preferably, the or each scanning surface is coated in an anti-reflective coating having an uneven finish.

Preferably, the or each scanning surface is coated in an anti-reflective coating having a mottled finish.

It will be appreciated that reference herein to "mottled surface" means a surface comprising uneven spots or blotches.

The coated material of the scanning surfaces of the present invention allows the device to view the text of a document to be scanned without causing significant aberration of the image, whilst significantly reducing the scattered light, which is used to illuminate the pages that are to be scanned. Furthermore, the material and coating reduces the glare at the captured image, which can lead to degraded contrast and so to further distortion of the imaged text.

Preferably, the scanning device comprises at least one two-dimensional image sensor.

By using a two-dimensional image sensor an accurate page image can be obtained without the requirement to move linear sensors using motorised or mechnical drive systems. The area sensors of the present invention would previoulsy have been thought to degrade the resolution of the page image obtained, such that the image would be unsuitable for Optical Character Recongition (OCR) processing. However, by obtaining multiple focussed images the processor of the present invention is configured to correct for deficiencies in the images obtained and a clear and accurate composite image is obtained, which is suitable for OCR. Preferably, the at least one image sensor comprises a pixelated image sensor, such as a complimentary metal-oxide semi-conductor (CMOS) sensor.

Preferably, the or each image sensor has a resolution of greater than about 5 megapixels.

Preferably, the or each image sensor is placed at the focus of a lens to form the imaging assembly. More preferably, wherein the relative positon of the lens and the sensor are pre-set.

Preferably, the moveable focussing housing comprises a voice coil motor mechanism.

Preferably, the lens is moveable by the movable focussing housing relative to the sensor to optimise the focal plane of the image.

More preferably, wherein the lens is independantly focusable.

Still more preferably, wherein the lens is movable to enable the scanning device to susbtantially simultaneously provide multiple clear images of sub-areas of two facing pages of a book into which the device is inserted.

Preferably, the imaging assembly is mounted on a printed circuit board.

Preferably, the optical axis of the or each lens intersects the scanning surface at an angle of less than about 25 degrees.

Preferably, the point of intersection of the optical axis and the scanning surface is offset from the mid-line of the scanning surface.

More preferably, the scanning device comprises three imaging assemblies aligned with each scanning surface. Preferably, the plurality of moveable focussing housings are each moveable to move the lens alone to scan substantially the entire area of each scanning surface.

Preferably, each of the plurality of moveable focussing housings are moveable to move the lens alone to each collect multiple images and each image is collected at each of two or more focal planes.

By scanning the entire scanning surface simultaneously, without requiring a motorised or mechanical drive system, the scanning device is able to capture an image of the entire page of the bound document. By collecting multiple images at different focal planes it is possible to 'stitch' the images together to form an accurate image of the entire page of a book adjacent to the scanning surface. The multiple images are collected by moving the moveable focussing housing without requiring movement of the entire imaging assembly along or across the full page.

More preferably, the plurality of imaging assemblies are arranged to obtain a plurality of focussed images across the entire area of both scanning surfaces of the scanning device.

Preferably, the scanning device comprises a plurality of imaging assemblies, wherein each moveable focussing housing is moveable to move the lens to collect multiple images from both scanning surfaces of the scanning device and wherein each image is collected at each of two or more focal planes.

More preferably, the scanning device comprises six imaging assemblies each having a moveable focussing housing, which is moveable to collect an image from a lens at each of two or more focal planes.

Preferably, each moveable focussing housing is focusable to collect an image from a lens at each of three focal planes. The configuration of the present invention moves the lens alone relative to the sensor to image substantially the full page of the document to be scanned, without requiring moviement of the entire imaging assembly along or across the page.

By simultaneously scanning the entire area of both scanning surfaces simultaneously, two opposing faces of a bound document can be read in real-time. This enhances a user's experience and makes use of the device more closely resemble reading of a book.

It is to be understood that substantially the entire area refers to a greater part. Preferably, substantially the entire area is more than about 50% of the surface area of the scanning surface/s. More preferably, substantially the entire area is more than about 70% of the surface area of the scanning surface/s. Still more preferably, substantially the entire area is more than about 80% of the surface area of the scanning surface/s, preferably more than about 90% of the surface area of the scanning surface/s, preferably more than about 95% of the surface area of the scanning surface/s.

Preferably, the or each imaging assembly is positioned beyond the outer edge of the scanning surface.

The outer edge of the scanning surface is understood to be the surface furthest from the spine of the bound document. By positioning the imaging assembly beyond the outer edge of the scanning surface, any risk of shadows being created on the opposing scanned page is avoided.

Preferably, the scanning device is surrounded by a light-tight enclosure.

A light-tight enclosure prevents unwanted light entering the device; for example, from external lighting, sunlight etc. and so prevents any unwanted distortion of the scanned image. Preferably, the scanning device further comprises one or more LEDs to illuminate the or each scanning surface.

By providing an array of LEDs, the illumination across the scanning surface, and so across the or each page to be scanned, can be controlled to minimise variation in illumination across the page that is being scanned and prevent any avoidable distortion of the scanned image.

More preferably, the or each LED is turned on when the scanning device is in a scanning mode.

By restricting use of the LED array to when illumination is required for scanning, the battery life of the device can be conserved.

Preferably, the scanning device is portable.

By providing a portable device, it is envisaged that the device could be used in the home, office, but also in restaurants to read menus etc. or on public transport.

Preferably, the scanning device further comprises a text to audio conversion means and an audio output for outputting audio data.

By converting the scanned, imaged data to audio output, a user can read aloud from a bound document by scanning through the pages of the document. The output is generated in real-time so that a visually-impaired user or a user with reading difficulties is able to experience the benefits and enjoyment of reading a book.

In a second aspect, the present invention provides a scanning method for scanning at least one page of a bound document comprising the steps of: inserting a wedge-shaped scanning device between two target pages of the bound document so that each of two scanning surfaces of the scanning device contacts a target page of the document;

collecting multiple focussed images of a target page sub-area ; combining the multiple collected target page sub-area images to produce a single composite image of the or each target page;

correcting the composite image to compensate for the angle of insertion of the wedge-shaped scanning device;

outputting a corrected composited image of at least one target page.

Preferably, the method comprises correcting the composite image to correct for the angled field of view of the wedge-shaped scanning device.

By combining the collected images the scanning method is able to compensate for keystone distortion. The scanning method avoids any distortion in the image caused by the different angles at which the wedge- shaped scanning device views the target page, i.e. because of a steeper viewing angle when the distortion between the near and far field images (furthest from the spine and nearest to the spine respectively) is greater.

Preferably, the scanning method simultaneously outputs an image of two opposing target pages of the bound document.

Preferably, the method comprises collecting multiple focussed images from three target page sub-areas.

Preferably, the method comprises collecting multiple focussed images wherein each of the multiple focussed images is collected at a different focal plane.

Preferably each image collected from the target page sub-area is trapezoidal.

The present invention processes the image using a "trapezium correction" to allow for perspective distortion and output an accurate image of the scanned page. Preferably, the images collected from the target page sub-areas are overlapping.

More preferably, the scanning method comprises the step of:

detecting one or more areas of overlap between the target page sub-areas; stitching together the sub-areas along the detected areas of overlap; and deleting any repetition in the or each area of overlap

Preferably, the scanning method further comprises the step of:

calibrating the range of movement of one or more movable focussing housings within the wedge-shaped scanning device.

The calibration step allows for accurate imaging whereby the possible range of movement of each movable focussing housing is known and can be pre-set according to the focal plane of a lens within the or each imaging assembly. The calibration to allow movement of the or each movable focussing housing allows the method to collect multiple images at different focal planes for each page of the book to be scanned.

More preferably, the scanning method further comprises the step of:

calibrating the illumination level of the scanning device.

Preferably, the step of calibrating the illumination level of the scanning device is carried out based on the or each imaging assembly viewing a blank white page.

Calibrating for illumination allows the method to compensate for illumination variation so that the un-distorted i.e. processed image generated by the scanning method will improve the contrast across the image that is output.

Preferably, the scanning method further comprises the step of converting image data into text data. More preferably, the scanning method further comprises the step of converting text data into audio data.

Preferably, the image of the target page is converted into text data using optical character recognition (OCR)

Preferably, audio data is output through a loudspeaker or similar audio output means.

For the purposes of clarity and a concise description, features are described herein as part of the same or separate embodiments; however it will be appreciated that the scope of the invention may include embodiments having combinations of all or some of the features described.

The invention will now be described by way of example with reference to the accompanying diagrammatic drawings, in which : -

Figure 1 is a perspective view of a scanning device constructed in accordance with the present invention;

Figure 2 is a view from the side of the wedge-shaped scanning device of Figure 1 ;

Figure 3 is a schematic cross-sectional view of the wedge-shaped scanning device of the present invention;

Figure 4 is a schematic plan view from above of one of the scanning surfaces of the scanning device;

Figure 5 is a schematic plan view from above of one of the scanning surfaces of the scanning device;

Figure 6 is a schematic plan view from above of one of the scanning surfaces of the scanning device; Figure 7 is a flow diagram showing the process of converting a scanned image file to an output text file; and

Figure 8 is a schematic illustration of the components of the scanning device of the present invention.

Referring to Figures 1, 2 and 3, the scanning device 1 of the present invention comprises a wedge-shaped device having two scanning/contact surfaces 3, 5 separated by an angle of between about 15 and 20 degrees. It is understood that "book" in the context of the present invention refers to a bound document having a set of printed sheets held together along a spine, optionally inside a cover. The "wedge-shaped" device 1 has a solid shape defined by two triangular end faces and three rectangular or trapezoid faces, two of which are the scanning surfaces 3, 5. By having a wedge-shape, the device can easily be inserted into a bound document, such as a book so that the scanning surfaces are in contact with or in close proximity to the page/s to be scanned. The wedge-shape of the device has a thicker end, which houses internal workings of the device including user interface components, and tapers to a leading edge 9, which is a thin edge that can be inserted near to the book's spine. This ensures that the full page width can be scanned by the scanning device 1. The scanning device can be a triangular prism and is shown in Figure 2 to have a thicker end, which is curved to allow a user to easily and comfortably grip and carry the device. The thicker end of the device 1 can also house a loudspeaker; other output means such as a headphone socket; and control means, such as buttons to control volume and playback.

The contact surfaces 3, 5 of the scanning device 1 are made from a transparent, clear plastic coated in an anti-reflective coating having a slightly mottled finish. The material and coating in combination allow the scanner 1 to scan the text of the book 7 without any significant aberration of the image generated, whilst reducing the scattered light which is used to illuminate the pages 7a, 7b that are to be scanned. The material and coating reduce the glare at the captured image, which would degrade contrast and lead to further distortion of the imaged text. The two contact surfaces 3, 5 meet at the leading edge 9 of the scanner 1.

Referring to Figure 3, the internal workings of the scanning device 1 include a complementary metal-oxide semi-conductor (CMOS) sensor 11, or other pixelated image sensor. The sensor 11 is a two-dimensional area sensor. The sensor 11 has a resolution of greater than 5 megapixels and is placed at the focus of a lens 13. The lens 13 and sensor 11 are held in position by a moveable, focusing housing 15. Each imaging assembly, comprising the two-dimensional sensor 11 ; a lens 13; and a movable focussing housing 15, is static; that is, the position of the imaging assembly is substantially fixed and is not required to move across the page to be scanned. However, the focussing housing 15 is movable to alter the relative position of the lens 13 with respect to the sensor 11. The movable focussing housing 15 moves the lens 13 alone to allow images to be collected at multiple focal planes. In the preferred embodiment, shown in Figure 3, the movement of the housing 15 to permit focusing is provided by a conventional voice coil motor (VCM) mechanism (not shown), such as known devices used for performing auto-focus on a camera phone or similar such device. The focussing housing 15 is required to move a distance of less than about 50 micrometres to achieve the required focussing, preferably moving less than about 10 micrometres or even a sub-micrometre distance to reach each of the required lens positions.

In alternative embodiments, the movement of the focusing housing 15 is provided by a squiggle motor, such as those provided by New Scale Technologies Inc. ; a MEMS actuated auto-focusing device, such as that provided by DXO; linear piezo actuators; a shape memory, alloy-actuated linear focusing mechanism, such as those supplied by Cambridge Mechatronics; or auto-focusing lenses that use an electronic signal to control the refractive index of an LCD lens, such as those supplied by Lens Vector or Varioptic, or any other type of linear drive for moving the lens 13 relative to the sensor 11. In contrast to the known use of these prior art devices for auto-focussing, this is not the use of the sensor and associated movement of the above-described lens arrangement 11, 13, 15 of the present invention. In the configuration of the present invention, the lens 13 is driven to each of a plurality of pre-determined positions relative to the sensor 11. That is, the relative positions of the lens 13 and the sensor 11 are set during the initial calibration of the scanning device 1. The movement mechanism of the focusing housing 15 enables the scanning device 1 of the present invention to simultaneously provide multiple clear images of sub-areas of the two facing pages 7a, 7b of a book 7 into which the scanning device 1 is inserted, by moving the lens 13 alone. The movable focusing housing 15 is driven to multiple predetermined positions so that the focus of the or each lens 13 is optimised for different zones within the field of view of the or each sensor 11. The lens arrangement 11, 13, 15 is configured so that each lens 13 is independently focusable during use to overcome resolution variation. The fixed, static array of two-dimensional sensors allows for multiple focussed images to be collected and a composite full-page image to be subsequently re-generated into a single corrected image suitable for future processing, such as OCR.

Figure 3 shows an example of a single imaging assembly 11, 13, 15 mounted on a printed circuit board 17. The imaging assembly 11, 13, 15 is such that the optical axis 13b of the lens 13 intersects the scanning surface 5, and so the target page 7b, at an angle of less than 25 degrees. The point of intersection between the optical axis 13b and the scanning surface 5 and the target page 7b is offset from the mid-line of the page by a pre-determined distance. The distance by which the optical axis 13b is offset is determined according to the focal length of the lens 13 and the range of the focussing mechanism, i.e. the movable housing 15.

An example embodiment is shown in Figure 3, wherein the scanning device 1 comprises a single imaging assembly 11, 13, 15. However, the scanning device is understood to have a plurality of imaging assemblies 11, 13, 15 arranged to scan the entire area of the facing pages 7a, 7b of the book 7 for a composite image of each of the pages 7a, 7b to be captured simultaneously. It is understood that the imaging assembly described with respect to Figures 4, 5, and 6 is also arranged to simultaneously look in the opposite direction to capture images from the facing page of the book.

Referring to Figures 3 and 4, in use, the pages 7 of the book that are to be scanned are separated by partially opening the book 7. The wedge- shaped scanning device 1 is then inserted between the pages 7a, 7b so that the leading edge 9 of the scanning device 1 is in contact with the inside spine 19 of the book 7. The insertion of the scanning device 1 into the book 7 allows the imaging assembly 11, 13, 15 to view the entire area of the text of both pages 7a, 7b.

The internal components 11, 13, 15, 17 of the scanner are surrounded by a light-tight enclosure, which secures the components in their correct positions and prevents any unwanted light entering the device; for example, from external lighting, sunlight etc. The pages 7a, 7b of the book 7 to be scanned are internally lit by a plurality of white light LEDs (not shown). The LEDs are controlled via the processor of the scanning device 1 and suitable drive circuitry. The LED array is multiplexed via pulse width modulation (PWM) in order to conserve the battery life of the scanning device 1. The LED array is controlled to turn on only when a user selects for scanning. The LED array is used with lenses (not shown) that maximise the collection efficiency of the light emitted to direct it into an elliptical pattern. This ensures that any variation in illumination on the page 7a, 7b to be scanned is minimised. It is envisaged that the calibration of the scanning device 1 will be carried out based on the imaging assembly viewing a standard white page so that white level illumination variation can be compensated for as required. The un- distorted i.e. processed image generated by the scanning device 1 will also compensate for any illumination variation and so improve the contrast across the image that is output. described with reference to Figures 4 and 5, each imaging assembly 20 used to collect multiple focussed images of a portion of the target page 7b, each at the optimum focus for the selected portion of the target page 7b, i.e. for the pre-determined sub-section of the field of view of the lens. As shown in Figure 4, each target page 7 is divided into three target page sub-areas 7' (indicated by dashed lines).

Referring to Figures 4 and 5, a plan view of the target page 7b and the text thereon is shown. Three imaging assemblies 20a, 20b, 20c are shown. It is understood that, in alternative embodiments of the invention, any number of imaging assemblies can be used with appropriate calibration being carried out in advance of scanning. In use, each of the sensor assemblies 20a, 20b, 20c views a target sub-area 22a, 22b, 22c of the target page 7b; for example, the target page 7b is divided into three target sub-areas 22a, 22b, 22c of equal size with a sensor assembly 20a, 20b, 20c allocated to each sub-area.

Referring to Figure 4, the optical axis 24 of the first imaging assembly 20a is shown to be along the centre line of the rectangular target sub-area 22a. Referring to Figure 5, each imaging assembly 20a, 20b, 20c is configured to view a trapezoidal target area 26. In use, the shortest edge of the trapezoidal target area 26 is closest to the spine of the book and the longest edge of the trapezoidal target area 26 is furthest from the spine of the book and closest to the imaging assembly 20.

As shown in Figures 3, 4, and 5, each imaging assembly 20a, 20b, 20c is located outside of, i.e. beyond the boundary of, the leading edge 7a of the book 7. This avoids any risk that the imaging assembly 20 will create a shadow on the opposing page of the book when the scanning device 1 is inserted between two pages 7a, 7b. The scanning device 1 is inserted once for simultaneously capturing an image of both opposing pages of the book 7a, 7b. The position of the imaging assembly 20 fixes the minimum object distance for each lens 13. The lens 13 is selected to match the optical resolution of the sensor pixels, i.e. 5 megapixels. The lens focal length is about 4mm, the paraxial magnification is about xl/40. The specification of the lens 13 used for the scanning device 1 is carefully selected to ensure that the size of the lens 13 is suitable for incorporating into a portable device, whilst providing a sufficiently accurate image for reading purposes.

Referring to Figure 5, it can be seen that each trapezoidal target area 22a, 22b, 22c overlaps with the adjacent target area 22 to a varying extent as a function of the proximity to the respective imaging assembly 20a, 20b, 20c, and so to the spine of the book. Thus, the target area 22a, 22b, 22c, i.e. the field of view of each imaging assembly 20, overlaps with the adjacent target area so that text in the area of overlap 26a, 26b is viewed, i.e. captured, simultaneously by two imaging assemblies 20a, 20b, 20c. Simultaneously imaging the overlapping area 26a, 26b allows for the image processor to compensate for keystone distortion and to "stitch" the separate imaged areas 26a, 26b together to produce an accurate overall image of each page.

With reference to Figure 3, the wedge-shaped scanning device 1 of the present invention views the object, i.e. the target page 7b of the book 7 at an angle. It has been found that the more oblique, i.e. steeper, the viewing angle, the greater the distortion between the near field images (furthest from the spine 19 of the book 7) and the far field images (nearest to the spine 19 of the book 7). This aberration known as keystone distortion is corrected for by the image processing software prior to presenting the image to OCR software. In addition to the keystone distortion, the clarity of the image, sometimes referred to as the focal quality, varies across the page from edge to spine. It is this variation that the moveable focusing mechanism 15 incorporated into the imaging assembly 20 corrects.

Referring to Figures 3 and 6, the image processing of the scanning device 1 is described in respect of the imaging assembly 20 and a voice coil motor (VCM) movement mechanism (not shown) used as the movable focusing housing 15. The range of movement of the focusing housing 15 is known, such that the upper and lower limits of the movement range can be calibrated with respect to the focal plane of the lens 13. For example, if the range of movement of the movable focusing housing 15 is 600 arbitrary units (a.u.) by careful calibration it is determined that for the chosen lens 13 having a focal plane 30, text closest to the edge of the spine of the book 7 is at best focus when the movable focusing housing 15 is moved to 200 a.u.. Half-way across the focal plane, i.e. around the mid-line 33 of the page of the book 7, the best focus is when the movable focusing housing 15 setting is 300 a.u.. For text furthest from the imaging assembly, i.e. towards the leading edge 9 of the scanning device 1, the movable focusing housing 15 setting is 400 a.u.. By carrying out careful calibration of each of the sensors 11 before scanning commences, each imaging arrangement 20a, 20b, 20c generates two or more images at different focal planes for each of the pages of the book 7.

In a preferred embodiment, shown in Figure 6, for a first selected target sub-area, three imaging assemblies 20a, 20b, 20c each take three images, with each image taken at a different one of three pre-determined focal distances 30, 33, 35 across the width of the page of the book 7. Nine separate images are produced per page and the image data is collated to re-create the full page, i.e. to produce a composite, full-page image. Multiple cameras and multiple focal distances provide multiple images, each having an area of overlap with the adjacent images. The re-creation of a single page of text can be implemented either during image processing or processing of the translated text. In a preferred embodiment of the present invention, the re-creation of a single page is implemented using the translated text; i.e. after the imaging data has been processed by optical character recognition (OCR) software.

The processor (not shown) uses optical character recognition (OCR) to splice together the highest contrast images and generate a full, decoded image of the page 7. In order to allow successful OCR i.e. for accurate image to text translation it is necessary to process the image that is to be imported. The OCR device sets a suitable resolution for the image that is imported. From testing carried out, it has been found that increasing the dots per inch (DPI) resolution of the image allows for a much improved accuracy of translation. When imported, the image has a DPI of about 96 and the processing tool increased this value to 300-600 DPI to meet the OCR requirement for accurate translation.

The correction for image/text overlap is achieved by "stitching" each of the collected images together. The area of overlap is recognised as a repetition of words or images between each collected image. The scanning device and method of the present invention searches for any repetition; for example for repetition of a defined number of words, to locate the area of image overlap. The stitching method thus locates the "seams" or overlaps where the multiple images need to be stitched together.

For example, when the re-creation of a single page is carried out after the OCR step of the scanning method outputs the text files, each line will add a "return" delimiter to signify or label the end of a line. The processor records the last word/s of the line before the delimiter, stores this word and then searches for this word in the next text file. When this word is located this indicates the point of overlap of the text image. The repeated words will then be deleted and the rest of the words will be "patched", i.e. appended, to form the full line of text from the page. The processor of the scanning device continues this process to stitch together each line horizontally, due to the page being split into sections according to the three separate focal planes 30, 33, 35.

The three imaging assemblies 20a, 20b, 20c are configured to image the full length of the page so that the full page can be re-created by stitching the image "patches" vertically in addition to the horizontal stitching. This is achieved using a similar process whereby the first word/s of each line are compared to the first word/s of the text file of the adjacent image until a match is made, i.e. until an overlap of word/s are detected. As before, when the overlap of the text image is detected all repeated words are deleted and the remainder of the text image is appended. This process is repeated for each line of the page, one line at a time, until the full page has been successfully stitched together and accurately re-formed. Referring to Figure 7, the image data collected from the plurality of imaging assemblies 20 is combined to produce a single composite image that is imported as an image file 40 to the microprocessor of the scanning device 1. The scanning device 1 sets the image resolution at step 42 and applies a pre-configured trapezium correction at step 44 to allow for the distortions arising due to collection of the image data from a wedge- shaped device, as previously described. The scanning device 1 applies brightness and contrast correction at step 46 before converting the processed image data to text at step 48. The scanning device then saves the output as a text file at step 50.

Referring to Figure 8, the scanning device 1 of the present invention comprises a white LED array 50 and LED drive circuitry 51 connected to a microprocessor 52. The LED array 50 and associated circuitry 51 are able to control LED illumination of the book to be read ; for example the scanning device 1 detects insertion of the device 1 into the pages of a book to switch on the LED array 50 to illuminate the pages of the book only when scanning is to be carried out, i.e. only when an image is to be captured. A user interface 53, such as a touch-screen, keyboard or other operating controls are also connected to the microprocessor 52 to allow a user to input data to or extract data from the microprocessor 52. For example, the interface 53 can recognise audio input or a button can allow a user to control volume or playback speed. The image capturing/camera components 54 also provide an input to the microprocessor 52 and comprises the sensor 11, lens 13, focussing housing 15 and printed circuit board 17, described with respect to Figures 1 to 6. The microprocessor 52 receives an input of the composite image captured from the plurality of imaging assemblies 20 and also controls processing of the image; for example to calibrate the scanning device 1 ; apply distortion correction and control contrast levels, as required. Adjustments of the brightness and contrast of the images are applied by the OCR processor. The OCR device is based on recognition of black and white characters and the imaged text thus needs to have a high contrast between the black and white characters on each imaged page. The greater the contrast between characters the greater the accuracy of the translation. It is envisaged that the adjustments are uniformly applied to all images imported into the OCR. By careful testing prior to use, appropriate adjustments and corrections are pre-set.

As previously described, the image processing of the scanning device 1 is based on a "trapezium correction" to allow for the perspective distortion created by the wedge-shape of the device and so the angle of the lens/sensor arrangement with respect to the page from which the image is to be captured. The acute angle at which the image is captured for each page and the trapezium correction that needs to be applied is a predefined correction applied to each book, because each image collected throughout a book is collected by the or each imaging assembly at the same angle. By careful testing the correction angle can be pre-set. The image processor ensures that the clearest possible image is output by applying distortion correction and adjusting contrast levels. The image is output to an optical character recognition (OCR) means within the microprocessor. The microprocessor carries out optical character recognition (OCR) to translate the image captured to text.

The text output from the microprocessor can be translated into any required format; for example, the text output is magnified and projected; or presented as audio commentary to a user, i.e. translated to speech for audio playback through an amplifier to a loudspeaker 57. The amplifier provides a volume control and the scanning device 1 further comprises a headphone-out socket through which audio output can be passed. The present invention also allows the scanned data to be stored 55 onto disc for later retrieval. The text file can be stored in a non-volatile memory. As a user moves the scanning device 1 through multiple pages of the bound document, the text files of each page will be automatically stored in sequential order; for a user to later play back the pages in the correct order as a visual or audio output, as required.

By ensuring that an accurate image at suitable resolution is output to the embedded OCR software, any inaccuracies in the text are avoided before they are passed to a projector or converted into audio commentary. It is envisaged that the scanning and conversion to text and speech is completed in real-time. The microprocessor 52 is connected to a storage means 55 from which and into which data can be extracted or stored. The storage means allows a user to both store and organise captured files. The scanning device of the present invention 1 further comprises a battery power source 59 coupled to a power management means 61. In addition to the functions previously described, the microprocessor 52 of the present invention also controls the image capture by the plurality of imaging assemblies; controls illumination of the LED array (the flash) and collates the overlapping images taken by the multiple imaging assemblies to re-create a collated image of the full page of the document.

The above described embodiment has been given by way of example only, and the skilled reader will naturally appreciate that many variations could be made thereto without departing from the scope of the claims. For example, the use of three sensors in the embodiments described above is for illustrative purposes only. A similar arrangement of lens and sensors to form imaging assemblies can be used having two, four, five, six or more imaging assemblies arranged along the scanning device 1. Furthermore, it is envisaged that in an alternative embodiment a single sensor having a wide angle lens is used.