This invention relates to a method of optically determining the speed of a surface by means of pic¬ ture rows successive in time, the picture rows being obtained from an array of photodetectors.

In a prior art method, the displacement of a surface during the time interval between the imaging of two picture rows successive in time is calculated by determining the correlation between the two pic¬ ture rows. The speed of the surface can been de¬ termined in this way. This correlation method re¬ quires that the surface to be imaged comprises dots relatively clearly distinguishable from the surround¬ ings, otherwise the speed cannot be determined re¬ liably. The method is not particularly applicable to the determination of the speed of surfaces yielding picture rows in which the intensities of the picture elements vary seemingly arbitrarily.

The object of the present invention is to pro¬ vide a method of determining the speed of nearly any surface which is patterned at least to some extent. Such a patterned surface may be, e.g., a wooden sur- face, a metal surface, a surface of a road, a cloth surface, or other similar surface yielding picture rows in which the intensities of the picture elements vary at least with sufficient magnification. Accord¬ ing to the invention, the determination of the speed _pf a surface comprises digitizing the analog signals from the array of photodetectors; storing the digit¬ ized picture rows successive in time, whereby the picture rows arranged side by side can be assumed to form a two-dimensional picture in the space-time domain; and determining the slope or a corresponding

quantity of a line drawn by each picture element in the picture row in said space-time domain to obtain a value proportional to the speed of the surface. Pre¬ ferably the determination of the value of the slope or a corresponding quantity of the line comprises the determination of gradients in the direction of the space-axis and the time-axis for each picture element in the picture row as well as the determination of the ratio of the space-axis and time-axis gradients of each picture element. The measuring resolution can be increased manyfold dy determining the average : value of the values calculated for the picture el¬ ements of each picture row. To improve the re¬ liability of the result it is possible, if desired, to select the picture elements to be averaged before the averaging operation, e.g., in such a manner that the value calculated for each picture element on the basis of its gradients is compared with a preset value. In this way picture elements which are hardly distinguishable from the surroundings will not affect the calculation result.

The basic idea of the method of the invention is that digitized picture rows obtained from the de¬ tector array are stored in a memory to form a picture in which picture elements having an intensity dif¬ ferent from that of the surroundings form lines the slope of which is proportional to the speed of the surface. According to the invention, the slope of* ^" the lines is measured digitally, e.g., by convoluting the .picture so as tor obtain space-axis and time-axis gradients, the slope being obtained as the ratio of these gradients. In this way the speed or a value proportional to it is obtained for each picture el¬ ement. As mentioned above, the measuring resolution can thereby be increased manyfold by averaging the

speed values of each picture row. The selection of the values to be averaged can be controlled by summing only those values the square sum of which exceeds a predetermined value. In this way the values of those picture elements only which are contained in some of the lines of the picture are averaged.

The method of the present invention is ap¬ plicable within a wide speed range and its measuring resolution is sufficient for the method to be applied to the measuring of the length of a body as well. In the method, the direction of the surface to be measured is also taken into account as well as the possible stopping of its movement. The method can be realized in parallel with image processing devices at high processing rate.

In the following the method of the invention will be described in greater detail with reference to the attached drawing, wherein

Figure 1 is a schematic view of a surface to be imaged and a photodetector array imaging it;

Figure 2 shows a two-dimensional picture formed according to the invention; and

Figure 3 shows the flow diagram of the method of the invention. Figure 1 shows schematically a moving surface 1 which is being imaged with a photodetector array 2. In practice, the photodetector array may be, e.g., a video camera, an array camera or even an array- of separate photodetectors. The photodetector array 2 images a row 3 of fpicture elements from the moving surface 1. The direction of movement of the surface 1 is indicated with a to-and-fro arrow V in Figure 1.

Figure 2 shows a two-dimensional picture in the space-time domain, formed by means of the arrangement of Figure 1. During imaging, the surface 1 of Figure

1 has moved to and fro in the horizontal plane of the figure. The picture of Figure 2 has been formed by superimposing successive picture rows from the de¬ tector array 2 one upon another in such a manner that a picture row obtained at a moment t is disposed at the upper edge of the picture, a picture row obtained somewhat later at a moment t + t is disposed below it, etc. As appears from Figure 2, picture elements with an intensity different from that of the sur- roundings define lines in the picture when they are displaced in the direction of movement with time (to the left at the top of Figure 2). According to the invention, the speed of the surface 1 is determined by determining the slope or a corresponding quantity of these lines, whereby the speed of the picture element can be easily calculated from this value.

Figure 3 shows by way of example a flow diagram of the realization of the method of the invention. The first step 4 of Figure 3 comprises imaging one picture row by means of the photodetector array 2. At the following step 5, this picture row is digitized so that each picture element in the picture row is given a digital value corresponding to its intensity. At the following steps 6 and 7, the digitized picture rows successive in time are stored by means of shift registers so that two entities comprising a desired number of picture rows are available in each par¬ ticular case. Both of these entities can be assumed to form the same portion of the picture of Figure 2. ^.t step 6, the ,time-axis gradient, or the y-axis gradient, is determined from this picture for each picture element. Correspondingly, the space-axis gradient, or the x-axis gradient, is determined for each picture element at step 7. These gradients can be determined, e.g., by means of so called convolu-

tion masks. Determination of the gradients of picture elements with convolution masks is described, e.g., in "Handbook of Pattern Recognition and Image Pro¬ cessing", Academic Press, Inc., 1986, ed. Tzay Y. Young and King-Sun Fu, p. 192-194, 216 and 217.

At step 8 the slope of the picture element is determined as the ratio dy/dx of the time gradient dy and the space gradient dx obtained from the steps 6 and 7 for each picture element. The data thus obtain- ed on the speed of the surface is already sufficient for certain applications. However, greater resolution is achieved by calculating the average value of the slopes of the picture elements in one and the same picture row. The summing operation is effected at step 9 and the averaging at step 10. At step 11, the square root of the square sum of the gradients dy and dx is determined for each picture element in order to be able to exclude from the slope values to be summed those which are not contained in any of the lines visible in the picture of Figure 2 and which there¬ fore mainly contain arbitrary information. At step 12, the square root of the square sum is then com¬ pared with a preset value. The calculation of the slopes to be summed at step 9 is controlled on the basis of the result of this comparison. In Figure 3, this is illustrated by the arrow between the steps 12 and 9. In this way only the slopes of picture el¬ ements contained in the lines are summed at step ^*" 9. Data on the number n of acceptable picture elements -s applied from .step 12 to step 10, which has also received data on the sum of acceptable slopes from step 9. Accordingly, the sum of the selected slopes Σ dy/dx can be divided by the number n of the summed slopes, thus obtaining the average value of the slopes of the line. This average value defines the

speed very accurately at the. imaging moment of one particular picture row.

The practical realization of the method of the invention through the steps of Figure 3, for instance, can be carried out by digital VLSI cir¬ cuits. So the A/D conversion step 5, for instance, can be realized with a TDC1048 circuit, manufacturer TRW Inc. The shift register and gradient calculation operations of steps 6 and 7 can be carried out, e.g., with a PDSP16488 circuit, manufacturer Plessey Semi¬ conductors. The slope calculation and the calculation of the square root of the square sum of the gradients in step 8 can be carried out with a Pythagorean pro¬ cessor of the type PDSP16330, manufacturer Plessey Semiconductors, whereby the other output is not the slope dy/dx but a corresponding angle or atan (dy/dx). The summing of the slope values at step 9 can be realized with a PDSP1601 circuit, manufacturer Plessey Semiconductors. The comparison operation and the counting of the number of acceptable values at step 12 can be realized with a comparator and a counter formed by conventional MSI components. Cal¬ culation of the final average value at step 10 can be carried out most simply by means of a microcomputer programmed for this purpose. The microcomputer there¬ by reads the sum of the slopes of the picture row and the number of the summed slope values and calculates their quotient, that is, the average value of * ^" the slopes of the line. This average value is directly .proportional to „the speed of the surface to be measure .

The method of the invention has been described above by way of example by means of one specific em¬ bodiment. So only one way of determining the slope or a corresponding quantity of a line drawn by each el-

ement in a picture row in the space-time domain has been disclosed. Being simple, this way is to be pre¬ ferred; however, it is to be understood that the slopes or corresponding quantities of lines occurring in a formed two-dimensional picture can be determined in other ways as well, for instance, mathematically or by image processing techniques.