Optical measurement system

Abstract

When measuring the speed of a moving object using a linear sensor camera, multiple linear sensor cameras are required, and when measuring with a single linear sensor camera, there are various constraints on the subject. [Solution] A linear sensor camera is positioned diagonally to the direction of movement of the object being measured. At predetermined time intervals, it captures patterns correlated in a direction perpendicular to the direction of movement of the object being measured. By using the image information separated by predetermined time intervals, the distance the object has traveled at predetermined time intervals is determined, and the velocity of the object is calculated.

Description

Technical Field

[0001] The present invention relates to a speed measurement system, and provides an optical imaging device using a one-dimensional image sensor (hereinafter referred to as a linear sensor), which is a system for obtaining the speed and direction of a moving object to be measured.

Background Art

[0002] Conventionally, in speed measurement systems for measuring the speed of a moving object, measurements using the Doppler effect, measurements using an optical sensor, measurements using a passing sensor, or measurements using an imaging device have been performed. As the imaging device, some use an area sensor, and some use a linear sensor.

[0003] In speed measurement using the Doppler effect, the sensor must be installed facing directly in the traveling direction of the measurement target. Also, in measurement using an optical sensor such as a laser displacement meter, the measurement range is narrow and the speed cannot be measured unless the trajectory of the measurement target is accurately positioned, and the device is expensive. In the case of a passing sensor, when the measurement target is long in the traveling direction and acceleration or deceleration occurs while passing through the sensor part, the speed cannot be accurately measured. Furthermore, in the case of an area sensor, operations such as performing matching processing of the measurement target between frames and calculating the speed based on the moving distance of the measurement target become complicated.

[0004] As a speed measurement system using a linear sensor, a method is disclosed in Patent Document 1 in which two linear sensor cameras are arranged on the extension lines of two reference lines, the imaging signals of the moving objects of each measurement target are recorded in one control device, and from the recorded imaging signals, the passing time of the reference line is measured and the speed of the moving object is calculated. Here, the pixel arrangement direction of the linear sensor is perpendicular to the traveling direction of the moving object.

[0005] Patent Document 2 describes a method for calculating the velocity of a sample by arranging two linear sensor cameras at a predetermined interval along the direction of travel of the object to be measured, sequentially recording the imaging signals from the two linear sensor cameras with their pixel array directions orthogonal to the direction of travel, and then calculating the time required to move the predetermined distance based on the amount of shift that maximizes the correlation coefficient while shifting the imaging signals from both cameras in the time axis direction. Patent Document 3 describes a method for calculating the falling velocity of feces using a similar optical system.

[0006] Patent Document 4 describes a method in which the direction of movement of a moving object and the pixel arrangement direction of a linear sensor are aligned in parallel, and the difference in height between the driver's cab and the cargo bed of a truck, where the pixel signal of the linear sensor changes significantly, is detected, the position of the height difference is tracked, and the speed of the truck is measured. In this method, the speed of only one truck passing directly beneath the linear sensor camera is measured.

[0007] Patent Document 5 describes a method that uses two linear sensors, positioned perpendicular and parallel to the direction of motion of a moving object, to calculate velocity and correct image linearity. In this method as well, the linear sensor camera measures only one moving object at a time.

[0008] Patent Document 6 describes a method for calculating the rope's speed by capturing the change in rope's thickness when the rope's thickness changes periodically, such as in an elevator wire rope, and the interval between changes is known. This is achieved by arranging the pixels of a linear sensor perpendicular to the rope's extension direction. This linear sensor camera is single, and only one moving object (wire rope) is measured. [Prior art documents] [Patent Documents]

[0009] [Patent Document 1] Patent No. 3434610 [Patent Document 2] Japanese Patent Publication No. 2000-162220 [Patent Document 3] Japanese Patent Publication No. 2023-143245 [Patent Document 4] Patent No. 6067504 [Patent Document 5] Japanese Patent Publication No. 11-88753 [Patent Document 6] Patent No. 7318395 [Overview of the Initiative] [Problems that the invention aims to solve]

[0010] Conventional speed measurement systems shown in Patent Documents 1, 2, and 3 require the placement of two linear sensor cameras with pixels arranged perpendicular to the direction of motion. Furthermore, in Patent Document 1, if there are multiple moving objects, some objects may overlap, making it impossible to measure their speed. Patent Document 2 does not consider the case of multiple moving objects.

[0011] Patent Document 4 describes a system with a single linear sensor camera whose pixels are arranged parallel to the movement of the moving object. However, only one object must be measured, and it must be directly below the linear sensor camera. This is suitable for wide subjects like trucks, but unsuitable for narrow subjects like bicycles.

[0012] In the conventional speed measurement system shown in Patent Document 5, it is necessary to arrange two linear sensor cameras with pixels in a direction perpendicular to the movement of the moving object and a direction parallel to the movement of the moving object. The moving object to be measured is one, and speed and linearity can be corrected.

[0013] Patent Document 6 describes a linear sensor camera with pixels arranged perpendicular to the movement of the moving object, but the measurement target is limited to only one object and is limited to moving objects with periodic shape characteristics.

[0014] Thus, when measuring the speed of a moving object using a linear sensor camera, multiple linear sensor cameras are required, and when measuring with a single linear sensor camera, there are various constraints imposed on the subject, posing a challenge. [Means for solving the problem]

[0015] In order to solve the above problems, when measuring the speed of a moving object with a linear sensor camera using a linear sensor, the linear sensor camera is arranged obliquely with respect to the traveling direction of the measurement target, the measurement target is imaged at a predetermined time interval, and the imaging information generated by the imaging device at the first and second imaging times at the predetermined time interval is used to obtain the distance traveled by the measurement target at the predetermined time interval and calculate the speed of the measurement target.

Effects of the Invention

[0016] According to the present invention, a speed measurement system for obtaining the speeds of a plurality of moving objects with one linear sensor camera is provided. This is a concept that does not exist in the conventional examples. Furthermore, when measuring the movements of a plurality of vehicles on a road, since they move on the plane of the road and do not overlap, the speeds of individual moving objects can be measured.

Brief Description of the Drawings

[0017] [Figure 1(a)] A diagram showing the structure of a linear sensor camera. [Figure 1(b)] A diagram showing a speed measurement system according to a first embodiment using a linear sensor camera. [Figure 2] A diagram for explaining the timing when measuring the speed of a vehicle with the linear sensor according to the first embodiment. [Figure 3] A diagram for explaining the output waveform of the linear sensor at each timing in FIG. 2. [Figure 4] Continuous timings in the driving cycle of the linear sensor of vehicle 7 in FIG. 1(b). [Figure 4-2] Continuous timings in the driving cycle of the linear sensor of vehicle 7' in FIG. 1(b). [Figure 5] Output waveform of the linear sensor at continuous timings. [Figure 5-2] Output waveform of the linear sensor at continuous timings. [Figure 6(a)] Conceptual diagram for measuring the speed of a vehicle with the linear sensor according to the first embodiment. [Figure 6(b)] Figure 6(a) shows the relationship between the distance traveled by the car and the distance traveled at the front end. [Figure 6(c)] Output waveform of the linear sensor at the timing shown in Figure 6(a). [Figure 6-2(a)] A conceptual diagram showing how a linear sensor of the first embodiment measures the speed of a fast-moving vehicle. [Figure 6-2(b)] Figure 6-2(a) shows the relationship between the distance traveled by the car and the distance traveled at the front end. [Figure 6-2(c)] Output waveform of the linear sensor at the timing shown in Figure 6-2(a). [Figure 6-3(a)] A conceptual diagram showing how a linear sensor of the first embodiment measures the speed of a slow-moving vehicle. [Figure 6-3(b)] Figure 6-3(a) shows the relationship between the distance traveled by the car and the distance traveled at the front end. [Figure 6-3(c)] Output waveform of the linear sensor at the timing shown in Figure 6-3(a). [Figure 7(a)] A detailed diagram illustrating how to measure the speed of the car shown in Figure 6(a). [Figure 7(b)] The output waveform of the linear sensor at the initial timing in Figure 7(a). [Figure 7(c)] This figure shows the transition of the pixel position at the edge of the image in Figure 7(a) over consecutive timings. [Figure 7(d)] This figure shows the amount of pixel position shift at consecutive timings as shown in Figure 7(c). [Figure 8(a)] An explanatory diagram of a speed measurement system according to a second embodiment of the present invention. [Figure 8(b)] Consecutive reading timings in Figure 8(a). [Figure 9(a)] Figure 8(b) shows the continuous drive timing of the moving body 10-1. [Figure 9(b)] This figure shows the output waveforms of the moving object 10-1 at consecutive timings. [Figure 9(c)] A diagram showing the pixel position changes at consecutive timings for a moving object 10⁻¹. [Figure 9(d)] A diagram showing the amount of pixel position shift at consecutive timings of a moving object 10⁻¹. [Figure 10(a)]Figure 8(b) shows the continuous drive timing of the moving body 10-2. [Figure 10(b)] This figure shows the output waveforms of the moving object 10-2 at consecutive timings. [Figure 10(c)] A diagram showing the pixel position changes at consecutive timings for a moving object 10⁻². [Figure 10(d)] A diagram showing the amount of pixel position shift at consecutive timings of a moving object 10⁻². [Figure 11(a)] Figure 8(b) shows the continuous drive timing of the moving body 10-3. [Figure 11(b)] This figure shows the output waveforms of the moving object 10-3 at consecutive timings. [Figure 11(c)] A diagram showing the pixel position changes at consecutive timings of a moving object 10⁻³. [Figure 11(d)] A diagram showing the amount of pixel position shift at consecutive timings of a moving object 10⁻³. [Figure 11-2(b)] This figure shows the output waveforms at different points in the motion of object 10-3 at consecutive timings. [Figure 11-2(c)] A diagram showing the change in pixel position at different locations during consecutive timings of a moving object 10-3. [Figure 11-2(d)] A diagram showing the amount of pixel position shift at different locations at consecutive timings of the moving object 10-3. [Figure 12(a)] An image in which the shape of a moving object has been reconstructed based on the speed calculated using the original number of changing pixels. [Figure 12(b)] An image in which the shape of a moving object has been reconstructed based on a speed calculated using an incorrect number of changing pixels. [Figure 12(c)] An image in which the shape of a moving object has been reconstructed based on a speed calculated using a different, incorrect number of changing pixels. [Figure 13(a)] An explanatory diagram of a speed measurement system according to a third embodiment of the present invention. [Figure 13(b)] Consecutive reading timings in Figure 13(a). [Figure 14(a)] Figure 13(b) shows the continuous drive timing of the moving body 10-4. [Figure 14(b)] This figure shows the output waveforms of the moving object 10-4 at consecutive timings. [Figure 14(c)] A diagram showing the pixel position changes at consecutive timings for a moving object 10⁻⁴. [Figure 14(d)] A diagram showing the amount of pixel position shift at consecutive timings of a moving object 10⁻⁴. [Figure 15(a)] Figure 13(b) shows the continuous drive timing of linear pattern 11-5. [Figure 15(b)] This figure shows the output waveforms at consecutive timings for linear pattern 11-5. [Figure 15(c)] This figure shows the pixel position transitions at consecutive timings in linear pattern 11-5. [Figure 15(d)] This figure shows the amount of pixel position shift at consecutive timings in linear pattern 11-5. [Figure 16(a)] Figure 13(b) shows the continuous drive timing of linear pattern 11-5. [Figure 16(b)] This figure shows the output waveform at consecutive timings of wave pattern 11-5'. [Figure 16(c)] This figure shows the pixel position transitions at consecutive timings in the wave pattern 11-5'. [Figure 16(d)] This figure shows the amount of pixel position shift at consecutive timings in the wave pattern 11-5'. [Figure 17(a)] Figure 13(b) shows the continuous drive timing of the shading pattern 11-5”. [Figure 17(b)] This figure shows the output waveform at consecutive timings of shading pattern 11-5". [Figure 17(c)] A diagram showing the pixel position transitions at consecutive timings in shading pattern 11-5". [Figure 17(d)] This figure shows the amount of pixel position shift at consecutive timings in shading pattern 11-5". [Figure 18(a)] A diagram showing the continuous drive timing of the black dot density pattern 11-5". [Figure 18(b)] This figure shows the output waveform at consecutive timings for the black dot density pattern 11-5". [Figure 18(c)]This figure shows the pixel position transitions at consecutive timings for the black dot density pattern 11-5". [Figure 18(d)] This figure shows the amount of pixel position shift at consecutive timings in the black dot density pattern 11-5". [Figure 19(a)] An explanatory diagram of a direction measurement system according to a fifth embodiment of the present invention. [Figure 19(b)] A diagram showing the direction to enter the toilet. [Figure 20(a)] A diagram showing the consecutive drive timings in which the moving bodies 15' and 15'' of the fifth embodiment are read. [Figure 20(b)] A figure showing the transition of the output waveform at consecutive timings in the fifth embodiment. [Figure 20(c)] A diagram showing the transition of the central position at consecutive timings in the fifth embodiment. [Figure 20(d)] A figure showing the progression of the pixel position shift amount at consecutive timings in the fifth embodiment. [Figure 21(a)] A diagram showing the continuous drive timings from which the moving body 10-3 of the sixth embodiment is read. [Figure 21(b)] A diagram showing the transition of the output waveform at consecutive timings in the sixth embodiment. [Figure 21(c)] A figure showing the difference pattern of the output waveform at consecutive timings in the sixth embodiment. [Figure 21(d)] A figure showing the transition of the pixel position shift amount at consecutive timings in the sixth embodiment. [Modes for carrying out the invention]

[0018] Figure 1(a) shows the cross-sectional structure of the linear sensor camera 1, in which the lens 3 and linear sensor 4 are housed inside the camera housing 2. The linear sensor camera 1 forms an image of the subject on the linear sensor 4 using the optical system of the lens 3, and sequentially reads out the charge generated by photoelectric conversion at each pixel 5 (not shown) of the linear sensor 4 to read the optical information of one line of the subject.

[0019] <First Embodiment> The concept of measuring the speed of a moving object with this linear sensor camera 1 is explained in Figure 1(b). The moving object in question is a car 7 traveling on a road 6. This first embodiment is one in which the shape of the moving object is known. For convenience, the image captured by the linear sensor 4 is shown corresponding to each pixel 5 of the linear sensor 4 that captures each point on the road. The dashed arrow in front of the car 7 (hereinafter abbreviated as "car") indicates the direction of travel. A distinctive feature here is that the linear sensor 4 is installed at an angle to the road. Therefore, the orientation of the pixels 5 is also at an angle to the direction of travel of the car. There is a center line 8 on the road 6. The car 7 is shown with different brightness levels on the hood 7-1 and the roof 7-2 for identification.

[0020] In the speed measurement system of the first embodiment of the present invention shown in Figure 1(b), the shape of the moving object (car on the road) is known, and the direction of travel is also determined. The moving object (car) has a pattern component (front grille at the tip of the hood) in a direction perpendicular to the direction of travel. This portion perpendicular to the direction of travel plays an important role in detecting the speed of the moving object.

[0021] In explaining the speed measurement system shown in Figure 1(b), a linear sensor camera 1 is fixed and reads optical information from a line of objects on the road 6, and a moving object (car 7) is passing through the object area of ​​this linear sensor camera 1. For the sake of explanation, the timings t1 to t6 of the linear sensor camera 1 readings correspond to which part of the moving car 7's optical information is being read, as shown by the dashed lines t1 to t6 in Figure 2.

[0022] Figures 3(a) to 3(f) show the output waveforms of the linear sensor 4 at timings t1 to t6, corresponding to the dashed lines t1 to t6 in Figure 2. In Figure 3(a), optical information from a portion of the car 7's hood 7-1 is superimposed on the optical information of the road 6 directly below the linear sensor camera and output. Optical information corresponding to the center line is also represented. In Figure 3(b), optical information from a portion of the car 7's hood 7-1 and roof 7-2 is output in a similar manner. Optical information corresponding to the stationary center line is represented each time because it is stationary. In Figure 3(c), optical information from the car 7's roof 7-2 and near the doors is output; in Figure 3(d), optical information from a portion of the car 7's rear hood 7-1 and roof 7-2 is output; in Figure 3(e), optical information from a portion of the car 7's rear hood 7-1 is output; and in Figure 3(f), only the road's optical information (including the center line) is output. Figure 3 shows the output waveform of the linear sensor 4 in the case where the optical information of car 7 is brighter than the optical information of road 6, but the following argument also holds true in the opposite case (when the car is black).

[0023] The above was a rough explanation of the speed measurement system shown in Figure 1(b) using the timings in Figure 2. In reality, the optical information of a line of an object on the road 6 is read during the drive cycle of the linear sensor 4 that constitutes the linear sensor camera 1. In the following explanation, this drive cycle will be referred to as Δt. This is also called the cycle time of the linear sensor 4. In Figure 4, the timing around t1 in Figure 2 is taken as t, and the subsequent drive timings t+Δt, t+2Δt, and t+3Δt correspond to this. Here, for the sake of explanation, the four dashed lines in Figure 4; t, t+Δt, t+2Δt, and t+3Δt indicate which part of the moving car 7's optical information is being read.

[0024] Although Figure 4 illustrates the timing around t1 for a car traveling on the right side of the diagram in Figure 2, the present invention can also be applied to a car traveling on the left side of the diagram in the opposite lane. This roughly corresponds to cars 7 and 7' at the optical information reading position of pixel 5 of the linear sensor 4 in Figure 1(b). For the sake of explanation, the same situation as the optical information of a part of the hood 7-1 of car 7, which is imaged at timing around t1 shown in Figure 4, is referred to as timing t, and the subsequent consecutive drive timings t+Δt, t+2Δt, and t+3Δt correspond to which part of the moving car 7'' is having its optical information read, as indicated by the four dashed lines in Figure 4-2: t, t+Δt, t+2Δt, and t+3Δt.

[0025] Figures 5(a) to 5(d) show the output waveforms of the linear sensor 4 at the dashed lines t, t+Δt, t+2Δt, and t+3Δt in Figure 4. In Figure 5(a), optical information from a portion of the hood 7-1 of the car 7 is superimposed on the optical information of the road 6 directly below the linear sensor camera and output. Figure 5(a) shows a portion of the output of the linear sensor. In Figure 5(b), the range of pixels imaged corresponding to the hood 7-1 of the car 7 is expanded. What is important here is that the pixel positions (the first pixels of the linear sensor) corresponding to the part of the hood closest to the side wall of the car (corresponding to the doors and side windows, but referred to as the hood for the sake of the drawing) do not change, while the pixel positions (the last pixels of the linear sensor corresponding to the hood) corresponding to the part of the hood closest to the front of the car shift backward. In Figures 5(a) and 5(b), the change in the number of pixels at Δt corresponding to the hood being imaged is denoted as the number of changed pixels, ΔLp. In Figure 5, the horizontal axis of the output waveform of the linear sensor corresponds to the pixel position; therefore, changes in the horizontal direction represent changes in the pixel position.

[0026] In Figure 5(c), the range of pixels imaged corresponding to the hood 7-1 of car 7 is further expanded. The initial position of the pixel output pattern corresponding to the hood remains the same as in Figure 5(b), but the final position is shifted further back. From Figure 5(b) to Figure 5(c), the change in the number of pixels corresponding to the hood at Δt is the same as ΔLp, assuming that the speed of car 7 remains the same. In the following explanation, we will assume that the speed does not change during speed measurement.

[0027] In Figure 5(d), the range of pixels imaged corresponding to the hood 7-1 of car 7 is further expanded. The initial position of the pixel output pattern corresponding to the hood remains the same as in Figure 5(c), but the final position is shifted further back. From Figure 5(c) to Figure 5(d), the number of pixels that change in Δt corresponding to the hood is the same as ΔLp.

[0028] Figures 5-2(a) to (d) show the output waveforms of the linear sensor 4 at the dashed lines t, t+Δt, t+2Δt, and t+3Δt in Figure 4-2. In Figure 5-2(a), optical information from a portion of the hood 7-1' of car 7' is superimposed on the optical information of the road 6 directly below the linear sensor camera and output. Figure 5-2(a) shows a portion of the output of the linear sensor. In Figure 5-2(b), the range of pixels imaged corresponding to the hood 7-1' of car 7' is expanded. What is important here is that the pixel positions corresponding to the part of the hood closer to the side wall of the car (the last pixels of the linear sensor) do not change, while the pixel positions corresponding to the part of the hood closer to the front of the car (the first pixels of the linear sensor corresponding to the hood) shift forward. In Figures 5-2(a) and 5-2(b), the change in the number of pixels at Δt corresponding to the hood being imaged is denoted as the change in the number of pixels, ΔLp'.

[0029] In Figure 5-2(c), the range of pixels imaged corresponding to the hood 7-1' of car 7' is further expanded. The final position of the pixel output pattern corresponding to the hood remains the same as in Figure 5-2(b), but the initial position is shifted further forward. From Figure 5-2(b) to Figure 5-2(c), the number of pixels that change in Δt corresponding to the hood is the same as ΔLp'.

[0030] In Figure 5-2(d), the range of pixels imaged corresponding to the hood 7-1' of car 7' is further expanded. The final position of the pixel output pattern corresponding to the hood remains the same as in Figure 5-2(c), but the initial position is shifted further forward. From Figure 5-2(c) to Figure 5-2(d), the number of pixels that change in Δt corresponding to the hood is the same as ΔLp'.

[0031] Figures 6(a) to 6(c) explain how to acquire the speed of car 7 in the cases of the dashed lines t, t+Δt, and t+2Δt in Figure 4. The dashed lines t, t+Δt, and t+2Δt in Figure 6(a) indicate which part of the moving car 7 is being read, corresponding to the reading timings t, t+Δt, and t+2Δt of the linear sensor camera 1, and are shown in the same way as in Figure 4(a). Let Δd be the actual distance the tip of the hood 7-1 travels during the drive cycle Δt of the linear sensor 4. On the other hand, let Δtd be the actual distance the position sensed by the linear sensor 4 moves along the tip of the hood 7-1 during Δt. Here, we consider the case where the linear sensor camera 1 is monitoring a car driving on the road from directly above. As the car moves, the timing shifts from t+Δt to t+2Δt, and the distance traveled is Δd, while the distance traveled by the sensing position is Δtd.

[0032] There is a relationship between the distance Δd traveled during Δt and the distance Δtd traveled along the tip of the hood 7-1, as shown in Figure 6(b). That is, if the direction of the road 6 and the direction of travel of the car 7 are the same, and the tip of the hood 7-1 of the car 7 is perpendicular to the direction of travel of the car 7, then for a linear sensor camera 1 positioned at an angle θ from the direction perpendicular to the direction of the road 6 (the pixel direction of the linear sensor is also tilted by θ), the relationship is expressed as Δd = Δdt·tanθ. The length ΔL of the hypotenuse, which corresponds to the longer side of the right triangle in Figure 6(b), corresponds to the width that extends over the sensing area Δt of the bonnet 7-1 observed by the linear sensor camera 1. This can also be expressed as Δd = ΔL·sinθ.

[0033] Figure 6(c) shows the output waveform of the linear sensor 4 at the dashed lines t and t+Δt in Figure 6(a). Similar to Figures 5(a) and (b), in Figure 6(c), the range of pixels imaged corresponding to the hood 7-1 of car 7 is expanded. The pixel positions at the beginning of the linear sensor corresponding to the hood on the side wall of the car do not change, while the pixel positions corresponding to the hood closer to the front of the car shift backward, and the number of pixels that change at Δt is denoted as ΔLp. This corresponds to the number of pixels that change at linear sensor 4, ΔLp, which is equivalent to the actual width ΔL that expands during Δt of the sensing area of ​​the hood 7-1 of the subject being observed by the linear sensor camera 1. The relationship between ΔL and ΔLp is determined by the optical system (magnification by the lens) of the linear sensor camera 1.

[0034] In Figure 6(a), the car's speed V is the distance Δd traveled by the tip of the hood 7-1 during the drive cycle Δt of the linear sensor 4, and is expressed as V = Δd / Δt. This becomes ΔL·sinθ / Δt, and the car's speed can be determined from the number of changing pixels ΔLp, which corresponds to ΔL.

[0035] When installing the linear sensor camera 1, which monitors vehicles traveling on the road as shown in Figure 1(b), the condition of observation from directly above the vehicle as shown in Figure 6(a) is only valid for some lanes. Therefore, when actually taking measurements, it is desirable to investigate the relationship between the pixels corresponding to each point on the road monitored by the linear sensor camera 1, the actual distance on the road, and the number of pixels on the sensor when installing the linear sensor camera 1. This is because the linear sensor camera 1 observes from an oblique direction except from directly below, and this correction is necessary. This is also true when using conventional area sensors.

[0036] Thus, by using the linear sensor camera 1 that monitors vehicles traveling on the road shown in Figure 1(b), it becomes possible to measure the speed of vehicles in multiple lanes by applying individual corrections. Furthermore, if the observation position of the linear sensor 1 is sufficiently above the road, lane-specific corrections may not be necessary. Furthermore, if it's not possible to monitor the lane directly below and observation is made from an oblique angle, it becomes necessary to correct for the tilt from the vertical.

[0037] The angle θ between the orthogonal direction of road 6 and the linear sensor camera 1 is determined by the relationship between the vehicle speed V and the drive cycle Δt of the linear sensor 4. The case when the vehicle speed is high is explained in Figures 6-2(a) to (c), and the case when the vehicle speed is low is explained in Figures 6-3(a) to (c). The reading timing of the linear sensor camera 1 is shown only as t and t+Δt, and the distance the position sensed by the linear sensor 4 moves along the tip of the hood 7-1 during Δt is denoted as Δtd in both cases. The concept is the same in both cases, so they will be explained simultaneously.

[0038] When the speed is high, the distance the tip of the bonnet 7-1 moves during the drive cycle Δt, as shown in Figure 6-2(a), is Δd', which is longer than Δtd. On the other hand, when the speed is low, as shown in Figure 6-3(a), the distance the tip of the bonnet 7-1 moves during Δt is Δd'', which is shorter than Δtd. In both cases, the distance the linear sensor 4 senses moves along the tip of the bonnet 7-1 during Δt is Δtd.

[0039] As shown in Figures 6-2(a) and 6-3(a), there is a relationship between the distances Δd' and Δd'' traveled during Δt and the distance Δtd traveled along the tip of the bonnet 7-1, as shown in Figures 6-2(b) and 6-3(b). That is, for the linear sensor camera 1, which is positioned at an angle θ' and θ'' from the direction perpendicular to the direction of the road 6, these can be expressed as Δd' = Δdt·tanθ' and Δd'' = Δdt·tanθ'', respectively. The lengths ΔL' and ΔL'' of the hypotenuses, which correspond to the longer sides of the right triangles in Figures 6-2(b) and 6-3(b), correspond to the width that extends over the sensing area Δt of the bonnet 7-1 observed by the linear sensor camera 1. These can also be expressed as Δd' = ΔL·sinθ' and Δd'' = ΔL·sinθ''.

[0040] Figures 6-2(a) and 6-3(a) show the output waveforms of the linear sensor 4 at dashed lines t and t+Δt, respectively, in Figures 6-2(a) and 6-3(a). Similar to Figure 6(c), in Figure 6-2(c), the range of pixels being imaged is expanded to correspond to the area near the door 7-3, the roof 7-2, and the hood 7-1 of the car 7, while in Figure 6-3(c), the range of pixels being imaged is expanded to correspond to the hood 7-1 of the car 7. The initial pixel positions of the linear sensor corresponding to the area near the door on the side wall of the car and the hood remain unchanged, while the pixel positions corresponding to the area on the hood closer to the front of the car shift backward. The number of pixels that change at Δt corresponds to Figures 6-2(c) and 6-3(c), and are denoted as ΔL'p and ΔL”p. These correspond to the number of pixels in the linear sensor 4, which corresponds to the width ΔL' and ΔL” that expands during Δt in the sensing area of ​​the hood 7-1 observed by the linear sensor camera 1.

[0041] In Figures 6-2(a) and 6-3(a), the car's speed V is the distance Δd traveled by the tip of the hood 7-1 during the drive cycle Δt of the linear sensor 4, and is expressed as V = Δd / Δt. Corresponding to Figures 6-2(c) and 6-3(c), the car's speed is V' = ΔL'·sinθ' / Δt and V” = ΔL”·sinθ” / Δt. When the drive cycle Δt determined by the linear sensor is constant, the accuracy of speed measurement can be improved by changing the angle of the linear sensor camera 1 relative to the road 6 according to the speed of the moving object being measured.

[0042] Let's actually input the values ​​and make an estimate in the case of Figure 6(a). As a premise, let's assume θ = 45°, Δt = 1 msec, and the car speed is 60 km / h. The distance traveled during Δt is Δd = 60000 / 3600 / 1000 = 0.016 m, and at the tip of the hood 7-1, Δdt = 0.016 m. ΔL = 0.022 m. The lane width on a general road is 3.5 m, and if we try to monitor the two lanes (upper and lower) with a line sensor camera tilted at 45°, the imaging area needs to be 3.5 × 2 × √2 = 9.9 m, and if we monitor with a 5000-pixel linear sensor, each pixel will be 2 mm. That is, in this case, the change in the number of pixels ΔLp in the drive cycle Δt is 11 pixels, which is sufficient for measurement. On a highway, we can increase θ further.

[0043] Figures 7(a) to 7(d) further explain the method for acquiring the motion velocity in the case of Figure 6(a). Figure 7(a) shows the timings t, t+Δt, t+2Δt, t+3Δt, t+4Δt, t+5Δt, t+6Δt, and t+7Δt with dashed lines. Figure 7(b) shows the output waveform of pixel 5 of linear sensor 4 at timing t. The output corresponds to the hood 7-1 of car 7. Figure 7(c) shows the transition of the pixel position at the edge of this hood pattern. The pixel position in Figure 7(c) corresponding to the output waveform at timing t is shown with a dashed line. Figure 7(c) shows the pixel positions at subsequent timings t+Δt to t+7Δt. In the figure, white circles correspond to rising edges (hoods closer to the side walls), and black circles correspond to falling edges (hoods closer to the front). The pixel position of the white circles does not change at timings after t+Δt. From t+Δt onward, the black circle moves along the front of the hood, and its pixel position changes linearly in response to the car's speed. The white and black circles correspond to the boundary regions of the moving object 7, and their changes are significant in response to the rising and falling edges of the pixel output pattern, corresponding to the extreme positions of brightness and darkness (positions where brightness and darkness change drastically).

[0044] Figure 7(d) shows the change in pixel position (number of changed pixels ΔLp) between timings t+Δt and t+7Δt. Timing t+Δt indicates the shift in pixel position between timings t and t+Δt, with a sign. Here, a + sign indicates an increase in pixel position, and a - sign indicates a decrease. White circles indicate that the pixel position remains unchanged at zero after timing t+Δt. Black circles indicate that at timings after t+Δt, the pixel position changes linearly in accordance with the car's speed, and the amount of shift is constant. This shows the shift in pixel position during Δt. This is the number of changed pixels ΔLp. In Figure 6(a), the car's speed V is the distance Δd traveled by the tip of the hood 7-1 during the drive cycle Δt of the linear sensor 4, and is expressed as V = Δd / Δt. This becomes ΔL·sinθ / Δt, and the car's speed can be determined from the number of changing pixels ΔLp, which corresponds to ΔL determined by the optical system. That is, if there is a region with a fixed number of changing pixels ΔLp as shown in Figure 7(d), the speed of a moving object can be determined.

[0045] As shown by the black circles in Figure 7(d), the velocity of the moving object can be determined by the pixel position shift amount, which is a constant, non-zero finite value. In this case, it corresponds to the falling edge of the pixel output waveform, which is near the front of the hood. In the following explanation, we will refer to points used to determine the velocity of a moving object in this way as feature points. In this case, the pixel output waveform falls at the location of the feature point, which corresponds to the extreme value of brightness. The selection of feature points is crucial when obtaining the velocity of movement, but when obtaining the direction of movement, it is sufficient to understand the direction in which the pixel output waveform corresponding to the subject shifts. This can be determined by the change in the width of the subject, and by the direction in which the midpoint of the subject's width shifts. The velocity is determined by the transition of the feature points.

[0046] The output waveform of the linear sensor shown in Figure 7(b) has the horizontal axis corresponding to the pixel position and the vertical axis corresponding to the pixel output at each position. Figure 7(c) shows the transition of the pixel position corresponding to the moving object at each timing, and Figure 7(d) shows the amount of shift in the pixel position at adjacent timings. In a linear sensor, this amount of shift corresponds to the number of pixels. Here, we use the English word "pixel" and refer to this amount of shift as the number of pixels changed ΔLp.

[0047] <Second Embodiment> In the first embodiment described above, the subject is a moving object such as a car 7 traveling on a road 6, where the direction of travel of the moving object is determined, the shape of the moving object is known, and the method for measuring speed has been explained for a moving object that has a component perpendicular to the direction of travel (direction of the road) (such as the front grille). If the direction of travel of the moving object is determined, and the linear sensor camera 1 is installed at an angle to the direction of travel, it is possible to determine a rough shape of the moving object even if its shape is unknown. Furthermore, if there is at least one linear region with a component orthogonal to the direction of travel (the expression has been changed because "direction of travel" would evoke the image of a car), the speed of the moving object can be detected. This method will be referred to as the second embodiment and will be described below.

[0048] A speed measurement system according to a second embodiment of the present invention will be described below. As an example of the second embodiment, as shown in Figure 8(a), there are moving bodies with a determined direction of travel, for example, moving bodies 10-1, 10-2, and 10-3 that travel at various speeds on lane 9. Although the shape of these moving bodies is unknown, they are moving bodies that have regions 11-1, 11-2, 11-3, and 11-3' that are perpendicular to the direction of travel. These moving bodies are imaged by a linear sensor 4 that is installed diagonally to the lane. The arrangement direction of the pixels 5 is also diagonal to the direction of travel of the moving bodies. This diagonal arrangement makes it possible to roughly grasp the shape of the moving body 10 and to grasp the speed of the moving body in the region perpendicular to the direction of travel. Each moving object, 10-1, 10-2, and 10-3, has only contour information and no pattern, and its brightness is assumed to be the same.

[0049] In explaining the speed measurement system shown in Figure 8(a), the linear sensor 4 reads optical information from one line of the moving object 10 on lane 9, and the moving object 10 is passing through the subject area of ​​the linear sensor camera 1. The timing of the linear sensor 4's readings corresponds to the location of the optical information being read from each moving body 10-1, 10-2, and 10-3, as shown by the dashed lines in Figure 8(b).

[0050] The features of the speed measurement system of the second embodiment, namely its ability to determine the shape of a moving object even when the shape is unknown, and to identify the location of the component perpendicular to the direction of travel, will be explained in detail using Figure 9 for a moving object 10-1 traveling on lane 9. The linear sensor 4 and its pixels 5, which are installed diagonally to lane 9, have the same optical system as in Figure 7(a).

[0051] In the speed measurement system shown in Figure 8(a), the linear sensor 4 reads optical information from one line of the moving object 10-1 on lane 9, as the moving object 10-1 passes through the subject area of ​​the linear sensor camera 1. The dashed line in Figure 8(b) indicates which part of the moving object 10-1's optical information is being read, corresponding to the timing of the linear sensor camera 1's reading. This is the same as in Figure 7(a). The pixel arrangement direction of the linear sensor 4 is tilted by an angle θ from the direction perpendicular to the direction of movement of the moving object 10-1. The moving object 10-1 has only contour information and no pattern, and its brightness is uniform.

[0052] Figure 9(a) shows a more detailed sequence of drive timings for detecting the moving object 10-1 shown in Figure 8(b). It shows the sequence of drive timings t+Δt, t+2Δt, t+3Δt, t+4Δt, t+5Δt, t+6Δt, and t+7Δt, starting from the initial detection timing t. The black circles in Figure 9(a) indicate which part of the contour information of the moving object 10-1 (shown by solid lines in the figure) is being read at each drive timing.

[0053] Figure 9(b) shows the output waveforms of the linear sensor 4 at each drive timing, corresponding to the moving object 10-1 shown in Figure 9(a). The output waveforms are shown from the timing t when the moving object 10-1 is first detected, to the consecutive drive timings t+Δt, t+2Δt, t+3Δt, and t+4Δt. Here, since the moving object 10-1 has only contour information and uniform brightness, the pattern of the output waveform of the linear sensor 4 changes at the contour portion of the moving object 10-1. The pixel positions corresponding to the rising and falling edges of this pattern correspond to the contour positions of the moving object read by the linear sensor 4.

[0054] Figures 9(a) and 9(b) show the method for acquiring the motion velocity, and Figures 9(c) and 9(d) provide a more detailed explanation. Figure 9(a) shows the reading locations of the motion 10-1 at timings t, t+Δt, t+2Δt, t+3Δt, t+4Δt, t+5Δt, t+6Δt, and t+7Δt, indicated by dashed lines. Figure 9(b) shows the output waveform of pixel 5 of the linear sensor 4 at each timing t, t+Δt, t+2Δt, t+3Δt, and t+4Δt. The output changes at the contour of the motion 10-1, and the pixel positions corresponding to the rising and falling edges indicate the contour information of the motion 10-1. Figure 9(c) shows the transition of the pixel position from timing t to t+7Δt. In the figure, white circles correspond to the rising edge (the rounded part near the side wall of the motion), and black circles correspond to the falling edge (the flat part at the beginning). The white and black circles correspond to the boundary regions of the moving object 10⁻¹, respectively, and represent the extreme values ​​of light and dark. The change in pixel position of the white circles becomes small from t+3Δt onwards. The black circles move along the leading flat portion from t+Δt to t+4Δt, and their pixel positions change linearly in accordance with the car's speed. A dashed line has been inserted in Figure 9(c).

[0055] Figure 9(d) shows the amount of pixel position shift at timings t+Δt to t+7Δt. Timing t+Δt indicates the shift in pixel position between timings t and t+Δt, with a sign. Here, a + sign indicates an increase in pixel position, and a - sign indicates a decrease. This is the same as in Figure 7(d). The white circles correspond to the small change in pixel position at timings t+3Δt and beyond, and are close to zero. The black circles correspond to the timings t+Δt to t+4Δ, where the pixel position changes linearly, and the amount of pixel position shift is constant, not zero. Corresponding to this, a dashed line is inserted in Figure 9(d), and the amount of pixel shift is denoted as ΔLp. This is also noted in Figure 9(b). The falling edge of the pixel output waveform in Figure 9(b), which corresponds to region 11-1 perpendicular to the direction of travel of the leading edge of the moving object 10-1 where the amount of pixel position shift is constant, is also a characteristic point where the amount of pixel position shift is constant, not zero. In this case as well, the pixel output waveform drops off at the feature points, corresponding to the extreme positions of brightness and darkness.

[0056] Similarly, for the detection of the moving body 10-2 shown in Figure 8(b), the timing of the drives is shown in detail in Figure 10(a). Since the component of the moving body 10-2 that is perpendicular to the direction of travel is the rear end, the timing of the start of detection of the rear end is t+2Δt, followed by the consecutive drive timings t+Δt, t+2Δt, t+3Δt, t+4Δt, t+5Δt, t+6Δt, and t+7Δt. The black circles in Figure 10(a) indicate which part of the contour information of the moving body 10-2 (shown by solid lines in the figure) is being read at each drive timing.

[0057] Figure 10(b) shows the output waveforms of the linear sensor 4 at each drive timing, corresponding to the moving object 10-2 shown in Figure 10(a). The output waveforms are shown for the consecutive drive timings t+Δt, t+2Δt, t+3Δt, and t+4Δt before and after the timing t+2Δt when the rear end of the moving object 10-2 is first detected. Here, as with moving object 10-1, the brightness of moving object 10-2 is uniform, with only contour information. The pixel positions corresponding to the rising and falling edges of the output waveform pattern correspond to the contour positions of the moving object read by the linear sensor 4. This is the same as in Figure 9(b). The white and black circles correspond to the boundary regions of the moving object 10-2, which is the subject, and represent the extreme values ​​of brightness and darkness, respectively. An example of a dynamic object with an appearance similar to that of moving object 10-2 is the lead car of a Shinkansen bullet train, where the front end is rounded and streamlined, and the rear end is the coupling section with the train, which is perpendicular to the direction of travel. Since it usually has multiple coupling sections, the speed can be determined for each passing train.

[0058] Figure 10(c) shows the change in pixel position from timing t to t+7Δt. In the figure, white circles correspond to the rising edge (the trailing edge of the moving object), and black circles correspond to the falling edge (the rounded part near the leading edge). The white circles move along the trailing edge from timing t+2Δt onward, so they change linearly. A dashed line has been inserted in Figure 10(c). At timing t to t+7Δt, the black circles move along the rounded part near the leading edge, so the pixel position does not change linearly, but gradually decreases in accordance with the roundness.

[0059] Figure 10(d) shows the pixel position shift amount at timings t+Δt to t+7Δt. The white circles indicate that the pixel position changes linearly from timing t+2Δt onward, and the pixel position shift amount becomes constant. Corresponding to this, a dashed straight line is inserted in Figure 10(d), and the pixel shift amount is denoted as ΔLp. This is also indicated in Figure 10(b). The rising edge of the pixel output waveform in Figure 10(b), which corresponds to the region 11-2 perpendicular to the direction of travel of the rear end of the moving object 10-2 where this pixel position shift amount becomes constant, also has a pixel position shift amount that is not zero but constant, and is a feature point. In this case, the pixel output waveform rises at the feature point, corresponding to the extreme value position of brightness or darkness. The amount of pixel position shift corresponding to the linear region perpendicular to the direction of motion becomes constant, as shown in Figures 9(d) and 10(d). The velocity can be calculated in this region. In the area corresponding to the rounded front of the moving object 10-2, as shown by the black circle in Figure 10(c), the change gradually decreases at the timing of t to t+7Δ, and the amount of pixel position shift approaches zero, as shown in Figure 10(d).

[0060] Similarly, for the detection of the moving body 10-3 shown in Figure 8(b), the timing of the drives is shown in detail in Figure 11(a). The moving body 10-3 has components perpendicular to the direction of travel at its front and rear ends. The detection of the front end and side wall (upper edge of the figure) begins at timing t, followed by a series of drive timings t+Δt, t+2Δt, t+3Δt, t+4Δt, t+5Δt, t+6Δt, and t+7Δt. On the other hand, the detection of the rear end and side wall (lower edge of the figure) follows a series of drive timings t', t'+Δt, t'+2Δt, t'+3Δt, and t'+4Δt. The black circles in Figure 11(a) indicate which part of the contour information of the moving body 10-3 (shown by solid lines in the figure) is being read at each drive timing.

[0061] Figure 11(b) shows the output waveforms of the linear sensor 4 at each drive timing corresponding to the detection of the front end and side wall (upper edge of the figure) of the moving object 10-3 shown in Figure 11(a). The output waveforms are shown at the consecutive drive timings t+Δt, t+2Δt, t+3Δt, and t+4Δt, starting from the timing t when the front end of the moving object 10-3 is first detected. Here, the brightness of the moving object 10-3 is assumed to be uniform, with only contour information. The pixel positions corresponding to the rising and falling edges of the output waveform pattern correspond to the contour positions of the moving object read by the linear sensor 4. The contour of this moving object 10-3 is entirely straight lines.

[0062] Figure 11(c) shows the transition of pixel positions at timing t to t+7Δt. In the figure, white circles correspond to rising edges (side walls of the moving object; upper part in the figure), and black circles correspond to falling edges (tip). The white and black circles correspond to the boundary regions of the moving object 10-3, which is the subject, and represent the extreme values ​​of brightness and darkness. Since the contour of the moving object 10-3 is entirely straight lines, the white circles change linearly, and the black circles also change linearly. In Figure 11(c), dashed lines have been inserted for both.

[0063] Figure 11(d) shows the amount of pixel position shift at timings t+Δt to t+7Δt. Both the white and black circles correspond to linear changes in pixel position, and the amount of pixel position shift is constant in both cases. Correspondingly, a dashed line is inserted in Figure 11(d), and the number of pixels that change is denoted as ΔLp1 and ΔLp2, respectively. This is also shown in Figure 11(b). If the width of the moving object 10-3 (the width of a car, for example) is further increased, there will be times when the rear end and front end become white circles and black circles, respectively. At those times, the transitions of the white and black circles indicating the pixel positions in Figure 11(c) will be parallel, and the change in the number of pixels ΔLp1 and ΔLp2 in Figure 11(d) will both be the same, meaning both will be feature points.

[0064] Similarly, since the other side walls of the moving body 10-3 (the lower part in the figure) are also straight lines, the black circles corresponding to the falling edge change linearly in pixel position, and the amount of pixel position shift is constant, as will be explained below. Figure 11-2(b) shows the output waveforms of the linear sensor 4 at each drive timing corresponding to the detection of the rear end and the side walls (lower part in the figure). The output waveforms at consecutive drive timings t', t'+Δt, t'+2Δt, t'+3Δt, and t'+4Δt, where the rear end of the moving body 10-3 is detected, are shown. The contour of this moving body 10-3 is entirely straight lines.

[0065] Figure 11-2(c) shows the transition of pixel positions at timing t'~t'+4Δt. In the figure, white circles correspond to the rising edge (rear end of the moving object), and black circles correspond to the falling edge (side wall; lower part in the figure). Since the contour of the moving object 10-3 is entirely straight, the white circles change linearly, and the black circles also change linearly. In Figure 11-2(c), dashed lines have been inserted for both.

[0066] Figure 11-2(d) shows the amount of pixel position shift at timing t'~t'+4Δt. Both the white and black circles correspond to a linear change in pixel position, and the amount of pixel position shift is constant in both cases. Correspondingly, a dashed line is inserted in Figure 11-2(d), and the number of pixels that change is denoted as ΔLp1' and ΔLp2'. This is also shown in Figure 11-2(b). Both the front and side walls (upper part in the figure) and the rear and side walls (lower part in the figure) of the moving object 10-3, where the pixel position shift amount is constant, have two non-zero constant pixel position shift amounts. In previous feature point cases, there was only one location where the pixel shift amount was constant and not zero, making feature point selection easy. However, in this case, it becomes unclear which location to adopt as the feature point in order to detect velocity.

[0067] As stated in Figure 7(b), if there is a region with a fixed number of changing pixels ΔLp, the velocity of a moving object can be determined. As shown in Figures 9(d) and 10(d) corresponding to moving objects 10-1 and 10-2, if there is one region with a fixed number of changing pixels ΔLp, the velocity of the moving object can be determined. However, if there are two regions with a fixed number of changing pixels ΔLp, such as moving object 10-3, the following determination is added to select the number of changing pixels ΔLp that corresponds to the velocity of the moving object.

[0068] The first criterion for determination is the sign of the number of changing pixels. In lane 9 of Figure 8(a), the movement is to the right. Therefore, as shown in Figures 9(d) and 10(d), the number of changing pixels ΔLp for determining the speed has a positive sign. Consequently, ΔLp1 in Figure 11(d) is negative and is therefore excluded, leaving only ΔLp2, which corresponds to the rear end. Therefore, the falling edge of the pixel output waveform in Figure 11(b), which corresponds to the region 11-3 perpendicular to the direction of travel of the tip of the moving body 10-3, is the feature point. The pixel output waveform falls at the location of the feature point, which corresponds to the extreme value position of brightness or darkness.

[0069] As shown in Figure 11-2(d), when both change pixel counts have a positive sign, the second criterion is a method of reconstructing and judging the shape of the moving object based on the velocity calculated using the two change pixel counts ΔLp1' and ΔLp2'. To explain using the moving object 10-3 in Figure 11(a), the interval between observation lines can be determined by the velocity calculated using the change pixel counts ΔLp1' and ΔLp2' in Figure 11-2(d). If the change pixel count ΔLp1' corresponding to the original velocity is used, the original moving object 10-3 is reconstructed as shown in Figure 12(a). However, if the incorrect change pixel count ΔLp2' is used, the shape formed by the black circles in Figure 12(b) is obtained. The change pixel count ΔLp2' in Figure 11-2(d) is less than half of the velocity obtained with ΔLp1', but for the sake of explanation, Figure 12(b) is shown as the case where the velocity is half of the original. Conversely, in the case where the speed is doubled, the shape will be formed by the black circles in Figure 12(c).

[0070] Using the wrong number of changing pixels ΔLp2' results in a reconstructed shape of the moving object, as shown by the black circles in Figure 12(b). In this reconstructed shape, the component perpendicular to the direction of motion that was present in the reconstructed image corresponding to the original speed in Figure 12(a) is absent. This allows us to determine which number of changing pixels to select. If the speed is faster than the original speed, the reconstructed image will look like Figure 12(c). Therefore, the rising edge of the pixel output waveform in Figure 11-2(b), which corresponds to the region 11-3' perpendicular to the direction of travel of the rear end of the moving body 10-3, is the feature point. The pixel output waveform rises at the feature point, corresponding to the extreme position of brightness or darkness.

[0071] Figure 11 shows the case where there are two fixed variable pixel counts ΔLp, but the same approach applies even if there are three or more. The region where fixed variable pixel counts ΔLp exist is where there is a linear component in the contour of the moving object. Furthermore, this linear component is not limited to the contour region of the outer periphery of the moving object; even if it is inside the moving object pattern, if it can be detected by the linear sensor camera 1, then the fixed variable pixel count ΔLp exists and the speed of the moving object can be determined. The choice of which speed to select can be determined by the shape of the reconstructed moving object. A useful method for determining the reconstructed moving object is to learn the shape of a moving object traveling on lane 9 and train the AI ​​accordingly. Since the shape of an object passing through a lane is usually fixed, it is efficient to train the AI ​​in advance and learn where to extract feature point patterns from the pixel output waveform of the object to determine the speed.

[0072] <Third Embodiment> In the first and second embodiments described above, we have shown that the velocity of a moving object can be detected if there is at least one linear region with a component orthogonal to the direction of motion. This linear region with an orthogonal component does not need to be strictly defined; it is sufficient if there is a pattern that correlates orthogonally to a degree that allows for velocity estimation. This will be referred to as the third embodiment and will be described below. The correlated patterns referred to here describe the distribution of fine regions with similar brightness within a moving subject that exhibits directionality. In addition to the straight lines of the moving subject's outline as described above, straight lines, wavy lines, and shading patterns extending perpendicular to the direction of motion within the subject, as described below, also fall under this correlated pattern region.

[0073] A speed measurement system according to a third embodiment of the present invention will be described below. As an example of the third embodiment, as shown in Figure 13(a), there are, for example, moving bodies 10-4 and 10-5 traveling at various speeds on lane 9. The contour shape of these moving bodies does not have a region perpendicular to the direction of travel, but rather has regions 11-4, 11-5, 11-5', and 11-5'' inside the contour of the moving body that are correlated in a direction perpendicular to the direction of travel. These moving bodies are imaged by a linear sensor 4 that is installed diagonally to the lane. The arrangement direction of the pixels 5 is also diagonal to the direction of travel of the moving bodies. The moving object 10-4 has only contour information and no pattern, and its brightness is the same. The contour of the moving object 10-5 does not have a linear region with a component perpendicular to the direction of motion, but within the moving object 10-5 there is a linear region 11-5 with a component perpendicular to the direction of motion as a pattern, as well as a wave-like pattern 11-5' that is close to a straight line and a shading pattern that extends in the vertical direction.

[0074] The regions 11-4, 11-5, 11-5', and 11-5'', which are correlated in a direction perpendicular to the direction of travel, are explained below. Region 11-4 corresponds to a contour pattern that is almost parallel to a straight line in a direction perpendicular to the direction of travel, and region 11-5 corresponds to a straight line pattern in a direction perpendicular to the direction of travel. Region 11-5' is a wavy region similar to a straight line pattern in a direction perpendicular to the direction of travel. Region 11-5'' is a shading pattern that extends in the vertical direction and is a pattern parallel to a straight line pattern in a direction perpendicular to the direction of travel.

[0075] When explaining the speed measurement system shown in Figure 13(a), the linear sensor 4 reads optical information from one line of moving bodies 10-4 and 10-5 on lane 9, and the dashed lines in Figure 13(b) indicate which part of each moving body 10-4 and 10-5 the optical information is being read from.

[0076] Figure 14(a) shows a more detailed view of the detection of the moving object 10-4 shown in Figure 13(b), in order of the driving timings. It shows the successive driving timings t+Δt, t+2Δt, t+3Δt, t+4Δt, t+5Δt, t+6Δt, and t+7Δt, starting from the initial detection timing t. The black circles in Figure 14(a) indicate which part of the contour information of the moving object 10-1 (shown by a solid line in the figure) is being read at each driving timing. Here, only successive driving timings are shown.

[0077] Figure 14(b) shows the output waveforms of the linear sensor 4 at each drive timing, corresponding to the moving object 10-4 shown in Figure 14(a). The output waveforms are shown from the timing t when the moving object 10-4 is first detected, to the consecutive drive timings t+Δt, t+2Δt, t+3Δt, and t+4Δt. Figure 14(c) shows the transition of the pixel position at the pattern edge of the moving object 10-4 from timing t to t+7Δt. In the figure, white circles correspond to rising edges and black circles correspond to falling edges.

[0078] Figure 14(d) shows the shift amount of the pixel position at timings t+Δt to t+7Δt. The contour of the moving object 10-4 is entirely curved. As shown in Figure 14(d), the white circles do not show any linear changes in Figure 14(c), but the black circles change linearly between t+2Δt and t+6Δt, as shown in Figure 14(c), and a constant ΔLp can be obtained as shown by the dashed straight line in Figure 14(d). From this, the velocity of the moving object 10-4 can be determined. This can also be averaged using adjacent pixel information by introducing the concept of moving averages, which will be explained later. Therefore, the falling edge of the pixel output waveform in Figure 14(b), corresponding to the region 11-4 perpendicular to the direction of motion near the tip of the moving object 10-4, is a feature point where the pixel position shift amount is not zero but constant. In this case, as shown in Figure 15(b), the pixel output waveform also falls at this feature point, corresponding to the extreme position of brightness or darkness.

[0079] Similarly, as a pattern within the moving body 10-5 shown in Figure 13(b), the case of a straight line pattern 11-5 in a direction perpendicular to the direction of travel is explained in Figures 15(a) to (d). In order of driving timing, Figure 15(a) shows the driving timings t, t+Δt, t+2Δt, t+3Δt, t+4Δt, and t+5Δt.

[0080] Figure 15(b) shows the output waveforms of the linear sensor 4 at each drive timing, corresponding to pattern 11-5 in the moving object 10-5 shown in Figure 15(a). The output waveforms are shown for consecutive drive timings t+Δt, t+2Δt, t+3Δt, and t+4Δt, starting from the timing t when pattern 11-5 is first detected. Here, as with moving object 10-4, the brightness of moving object 10-5 is uniform, using only contour information. The pixel positions corresponding to the rising and falling edges of the pattern in the output waveform correspond to the contour positions of moving object 10-5 read by the linear sensor 4. In Figure 15(b), there is further a dip in the output waveform corresponding to pattern 11-5 within moving object 10-5, which shifts backward along with the drive timing.

[0081] Figure 15(c) shows the transition of pixel positions at the point of change in the output waveform from timing t to t+5Δt. In the figure, white circles correspond to the rising edge (side wall of the moving object), and black circles correspond to the falling edge (opposite side wall of the moving object). Triangles correspond to the position of pattern 11-5. The positions of the white and black circles remain unchanged, but the triangles shift backward with the timing.

[0082] Figure 15(d) shows the amount of pixel position shift at timings t+Δt to t+5Δt. The white and black circles remain unchanged at zero, but the triangular pixel position shift amount ΔLp corresponding to pattern 11-5 is not zero but constant. From this, the velocity of the moving object 10-5 can be calculated. Therefore, the dip in the pixel output in Figure 15(b), which corresponds to the region pattern 11-5 perpendicular to the direction of motion of the moving object 10-5, is a characteristic point. In this case, there is also a dip in the pixel output waveform at this characteristic point, which corresponds to the extreme value position of brightness or darkness.

[0083] Similarly, as a pattern within the moving body 10-5 shown in Figure 13(b), the case of a wave-like region 11-5' similar to a linear pattern in a direction perpendicular to the direction of travel is explained in Figures 16(a) to (d). In order of driving timing, Figure 16(a) shows the driving timings t, t+Δt, t+2Δt, t+3Δt, t+4Δt, and t+5Δt.

[0084] Figure 16(b) shows the output waveforms of the linear sensor 4 at each drive timing, corresponding to the wave pattern 11-5' in the moving body 10-5 shown in Figure 16(a). The output waveforms at consecutive drive timings t+Δt, t+2Δt, t+3Δt, and t+4Δt after the timing t at which pattern 11-5' is first detected are shown. The white and black circles are the same as in Figure 15(b), but corresponding to the wave pattern 11-5' in the moving body 10-5, there is a dip in the output waveform, and it shifts backward along with the drive timing.

[0085] Figure 16(c) shows the transition of pixel positions at the point of change of the output waveform at timing t to t+5Δt. The white and black circles in the figure are the same as in Figure 15(c). The triangles correspond to the position of the wave pattern 11-5' and, as in Figure 15(c), do not form a straight line but fluctuate in a wave-like manner. The positions of the white and black circles do not change, but the triangles shift backward along the straight line in Figure 15(c) as the drive timing progresses, undulating along the line.

[0086] Figure 16(d) shows the amount of pixel position shift at timings t+Δt to t+5Δt. The white and black circles remain unchanged at zero, but the triangular pixel position shift amount ΔLp corresponding to the wavy pattern 11-5' fluctuates in a wave-like manner, as illustrated by the triangle along a constant line in Figure 16(d). In this case of wave-like fluctuation, the moving average of the pixel positions (the average of the positions of multiple adjacent preceding and succeeding pixels) results in the dashed line (straight line) shown in Figure 16(c). The amount of pixel position shift also becomes a constant ΔLp, as shown by the dashed line at the triangular point in Figure 16(d). By calculating the pixel position shift amount ΔLp, the velocity of the moving object 10-5 can be calculated. Even when taking the moving average of the pixel positions for the non-wavy straight line pattern 11-5, the amount of pixel position shift remains a constant ΔLp, as shown by the dashed line at the triangular point in Figure 15(d). Therefore, the characteristic point is the dip in the pixel output shown in Figure 16(b), which corresponds to the region pattern 11-5' perpendicular to the direction of motion of the moving object 10-5. At this characteristic point, there is also a dip in the pixel output waveform as shown in Figure 16(b), which corresponds to the extreme value position of brightness or darkness.

[0087] Similarly, as a pattern in the moving body 10-5 shown in Figure 13(b), the case of shading pattern 11-5" which has light and dark shading in the direction of motion and extends linearly in a direction perpendicular to the direction of motion is explained in Figures 17(a) to (d). In order of the driving timing, Figure 17(a) shows the driving timings t, t+Δt, t+2Δt, t+3Δt, t+4Δt, and t+5Δt.

[0088] Figure 17(b) shows the output waveforms of the linear sensor 4 at each drive timing, corresponding to the shading pattern 11-5" in the moving object 10-5 shown in Figure 17(a). The output waveforms are shown for consecutive drive timings t+Δt, t+2Δt, t+3Δt, t+4Δt, and t+5Δt, starting from the timing t when the pattern 11-5" is first detected. The white and black circles are the same as in Figure 15(b), but corresponding to the shading pattern 11-5" in the moving object 10-5, there is a dip in the output waveform, which shifts backward with the drive timing. The extreme value (bottom in the figure) of this dip in the output waveform is called a feature point. The shading pattern 11-5'' in the moving object 10-5 shown in Figure 17(a) was a pattern where the center was dark, but the same argument holds true for shading patterns where the periphery is dark and the center is bright. However, in this case, it is a rise in the output waveform, and the extreme (top) position of the rise in the output waveform becomes the feature point.

[0089] Figure 17(c) shows the transition of pixel positions at the point of change in the output waveform from timing t to t+6Δt. The white and black circles in the figure are the same as in Figure 15(c). The triangles correspond to the positions of feature points in the shading pattern 11-5”. A feature point here is the pixel with the smallest output in the shading pattern 11-5”. In Figure 17(c), the positions of the white and black circles do not change, but after the drive timing t+3Δt, once the darkest part of the shading pattern is recognized, the triangles shift backward along the straight line in Figure 17(c) in accordance with the drive timing.

[0090] Figure 17(d) shows the amount of pixel position shift at timings t+Δt to t+6Δt. The white and black circles remain unchanged at zero, and the triangles are also zero up to the drive timing t+2Δt, before the darkest part of the shading pattern is recognized. However, at the drive timings t+4Δt, t+5Δt, and t+6Δt, after the darkest part is recognized, the pixel position shift amount ΔLp of the triangles becomes a constant line as shown in Figure 17(d). By calculating the pixel position shift amount ΔLp in this region, the velocity of the moving object 10⁻⁵ can be calculated. Therefore, the characteristic points are the dips in the pixel output in Figure 17(b) that correspond to the region pattern 11-5'' perpendicular to the direction of motion of the moving object 10-5. At these characteristic points, there is also a dip (minimum value) in the pixel output waveform, which corresponds to the extreme values ​​of brightness and darkness. Alternatively, a moving average can be obtained using the shading pattern 11-5'', and similarly, the dips (minimum values) will be characteristic points.

[0091] In the third embodiment, it was sufficient for the correlation to be in the direction orthogonal to the direction of travel. Figure 18(a) shows the case of a black dot density pattern 11-5"" which is composed of multiple black dots and distributed in a Gaussian waveform. In this case, the output waveform of the linear sensor 4 corresponding to each black dot has spike-like dips in the pixel output. However, this makes it difficult to identify feature points, so a moving average of the pixel positions of the dips (average of the positions of multiple adjacent pixels above, below, left, and right) is obtained, and a graph of the pixel position dependence of the density of dips is obtained and arranged in order of driving timing, resulting in Figure 18(b).

[0092] The shift of the extreme value (bottom) of the depression, which is the pixel position dependent on the density of the depression, can be determined as shown by the triangle in Figure 18(c), and the pixel position shift amount ΔLp of the triangle becomes a constant line as shown in Figure 18(d). Even with this, the pixel position shift amount ΔLp can be calculated, and the velocity of the moving object 10⁻⁵" can be calculated. In this way, there is also a method to obtain the velocity by adding one step (moving average acquisition) from the output waveform to determine the transition of feature points. Therefore, the characteristic point is the bottom of the dip in the moving average obtained from the pixel output in Figure 18(b), which corresponds to the case of the black dot density pattern 11-5"" perpendicular to the direction of motion of the moving object 10-5. At this characteristic point, there is also a dip (bottom of the distribution) in the pixel output waveform, which corresponds to the extreme value position of brightness and darkness.

[0093] The correlation in the direction perpendicular to the direction of progression, as referred to here, involves focusing on a single black dot and determining the direction of correlation by looking at the direction (the direction of the density distribution) of the nearest adjacent black dot. This concept is the same for line segments, shading, and black dot density. It becomes clearer when a moving average is obtained. This concept of moving averages can be introduced as a method for handling other cases.

[0094] In the explanations for Figures 14-18, there was only one case where the fixed number of changing pixels ΔLp was constant. However, as shown in Figure 11, if there are multiple regions with a fixed number of changing pixels ΔLp, the selection can be made by reconstructing and judging the shape of the moving object based on the velocity calculated using multiple numbers of changing pixels.

[0095] <Fourth Embodiment> In the first, second, and third embodiments described above, we have explained a method for determining the velocity of a moving object, assuming that there is a pattern within the object that correlates with the direction of motion and perpendicular to it. However, when measuring the velocity of feces, for example, as in Patent Document 3, there may be cases where such a pattern does not exist in the feces. In such cases, it is necessary to draw a pattern (straight line) perpendicular to the direction of motion (fall) of the moving object before observing it with an optical system.

[0096] Methods for drawing patterns (straight lines) in a perpendicular direction include mechanically scratching the surface (e.g., with a high-pressure water jet), optically scratching it using a laser like a laser marker, and using a chemical reaction by rapidly spraying a chemical solution in a linear fashion. In order to draw a pattern perpendicular to the direction of travel, the drawing speed must be sufficiently high relative to the speed of movement.

[0097] <Fifth Embodiment> In the first, second, third, and fourth embodiments described above, we have explained methods for obtaining the velocity of a moving object. However, by using the optical system of the present invention, it is possible to determine not only the velocity but also the direction. This will be explained below in the fifth embodiment.

[0098] Regarding the direction of motion, as already shown in Figure 1(b), for car 7 moving to the right on the road, the pixel position corresponding to the part of the hood closest to the front of the car (the last pixel of the linear sensor corresponding to the hood) shifts backward, as shown in Figures 4 and 5. On the other hand, for car 7' moving to the left on the road, the pixel position corresponding to the part of the hood closest to the front of the car (the first pixel of the linear sensor corresponding to the hood) shifts forward, as shown in Figures 4-2 and 5-2. Consequently, the pixel position shift at the center point of the car pattern also shifts in different directions, backward and forward, depending on whether the car is moving to the right or left.

[0099] As another fifth embodiment of the present invention, a toilet inside a building will be described as an example. In Figure 19(a), the toilet 12 consists of five individual stalls (12-1 to 12-5). Each stall is equipped with a toilet bowl 13-1 to 13-5, and doors 14-1 to 14-5 are installed at the entrance to each stall. To determine when people enter and exit this toilet, the area near the doors is monitored by a linear sensor camera 1. The image captured by the linear sensor 4 is shown corresponding to each pixel 5 of the linear sensor 4 that captures each point in front of the toilet. A distinctive feature here is that the linear sensor 4 is installed at an angle to the extension direction of the entrance. Since the extension direction of the entrance is perpendicular to the direction of entry and exit, the angle θ used so far is as shown in Figure 19(a).

[0100] In the case of the moving object 15 in Figure 19(b), which is similar to pattern 11-5 in Figure 15(a), when entering the toilet (arrow direction), the drop in the output waveform coincides with the drive timing, and the pixel position shifts backward, as shown in Figures 15(b) and (c). Conversely, when leaving the toilet, the pixel position shifts forward. Next, if the moving object is a person, their orientation is indicated by the direction of their noses, specifically for people entering 15' and people leaving 15''. When the moving object is a person, they enter and exit the toilet facing forward (assuming no one enters the toilet sideways or at an angle), so the line connecting both shoulders is perpendicular to the direction of movement, as shown in Figure 15. The black circles in Figure 19(b) represent a head with black hair, and this person's entry and exit from the toilet entrance to each individual stall is observed by a linear sensor. In Figure 19(a), this is represented by corresponding the pixel 5 positions of the linear sensor 4, which observes each part of the toilet entrance using a lens system.

[0101] Figure 20(a) shows the entry and exit of toilet users. Person 15 entering is shown using stall 12-2, and person leaving is shown coming out of stall 12-3. The timing t corresponds to the consecutive drive timings t, t+Δt, t+2Δt, t+3Δt, t+4Δt, and t+5Δt of the linear sensor 4, and Figure 20(a) shows which part of the person entering or leaving the toilet the optical information is being read from. This is the same as explained earlier.

[0102] For ease of explanation, Figure 20(a) shows diagrams corresponding to people 15 and 15' entering and exiting. Figure 19(b) shows the output waveforms of the linear sensor 4 at each drive timing. It shows the output waveforms at consecutive drive timings t+Δt, t+2Δt, t+3Δt, t+4Δt, and t+5Δt, starting from the timing t when the entry of person 15 is first detected. On the other hand, the drive timing for exiting 15' is in the order of t+5Δt, t+4Δt, t+3Δt, t+2Δt, t+Δt, and t. The drive timing is reversed for entering and exiting, and this is distinguished in Figure 20(b) by dashed arrows and dotted arrows. To avoid confusion, Figure 20(b) is further labeled with t', t'+Δt, t'+2Δt, t'+3Δt, t'+4Δt, and t'+5Δt.

[0103] The characteristic point of the output waveform at timings t to t+5Δt shown in Figure 20(b) is the central position of the output waveform of the contour corresponding to the outline of the human body 15, 15'. In Figure 20(c), the central position of the human body 15 attempting to enter the toilet stall 12-2 at driving timings t+Δt, t+2Δt, t+3Δt, t+4Δt, and t+5Δt is indicated by a triangle. Here, the pixel position corresponding to the central position shifts backward as shown by the dashed arrow. On the other hand, the pixel position corresponding to the central position of the human body 15' emerging from toilet stall 12-3 shifts forward, as indicated by the dashed arrow. In the first to third embodiments of the present invention, the rising, falling, and dropping points of the pixel output served as feature points for obtaining velocity, corresponding to the extreme positions of brightness and darkness. However, in embodiment 5 of the application for determining direction, the central position of the moving object, rather than its outline, is more suitable as a feature point.

[0104] Figure 20(d) shows the pixel positions of feature points corresponding to human body 15 and 15' at timings t+Δt to t+5Δt. Human body 15 shifts backward at the driving timing, and human body 15' shifts forward at the driving timing. In this way, the shift in the pixel positions corresponding to the feature points reveals information about movement in and out, i.e., the direction of movement.

[0105] While systems that use sensors in each individual toilet stall to announce toilet occupancy status are already in practical use, they require the installation of sensors in each toilet stall and the management of information from each individual sensor. Furthermore, the idea of ​​monitoring a toilet space with area sensors is visually unappealing and met with resistance. Linear sensors, on the other hand, can monitor entry and exit simultaneously, and since stationary objects don't appear in the image, this reduces the perceived resistance.

[0106] <Sixth Embodiment> In the above explanation, the pixel output waveforms at adjacent drive timings were compared, and the motion speed and direction of the motion were determined from the changes therein, but in the present invention, adjacent At the drive timing It is also possible to insert a process that acquires the difference in pixel output waveforms and removes the background still image to reduce the amount of information. This method will be explained using Figures 21(a) to (d) in the case described in Figures 11(a) to (d). The drive timing for detecting the moving object 10-3 shown in Figure 21(a) is the same as in Figure 11(a). The output waveforms of the linear sensor 4 at each drive timing corresponding to the detection of the front end and side wall of the moving object 10-3 are shown in Figure 21(b). Here, when the difference in pixel output waveforms at adjacent drive timings is acquired, as shown in Figure 21(c), the difference output pattern becomes an output pattern with widths ΔLp1 and ΔLp2, where the width corresponds to the pixel position shift amount, and the change in width is a constant width as shown in Figure 21(d). The subsequent processing is the same as in Figure 11.

[0107] <Relationship between claims, embodiments, and corresponding drawings> Claim 1... corresponds to the first, second, third, fourth, fifth, and sixth embodiments of the present invention. Claim 2... In particular, it corresponds to Figures 5, 7, 9, 10, 11, 14, and 15. Claim 3... In particular, it corresponds to Figures 17(b) and 18(b). Claim 4... In particular, it corresponds to Figures 18(a) and 18(b). Claim 5... In particular, it corresponds to Figures 16(c) and 16(d). Claim 6... In particular, it corresponds to Figures 21(c) and 21(d). Claim 7... In particular, it corresponds to Figures 11(d), 15(d), 16(d), 17(d), and 21(d). Claim 8... In particular, it corresponds to Figures 12(b) and 12(c). Claim 9...A fourth embodiment of the present invention. [Industrial applicability]

[0108] The present invention provides a simple system for measuring the speed and direction of moving objects using a single linear sensor camera. There are no conventional methods for measuring the speed of multiple moving objects with a single camera, and this system is easy to maintain and has high industrial applicability. [Explanation of Symbols]

[0109] 1 Linear sensor camera 2 Camera housing 3 lenses 4 Linear Sensors 4" and 4"' area sensors 5 pixels 6 road 7. Automobile (car) 7-1 Hood 7-2 Roof Near door 7-3 8 Center Line 9 lanes 10-1, 10-2, 10-3, 10-4, 10-5 moving objects 11-1, 11-2, 11-3, 11-4, 11-5 Regions perpendicular to the direction of travel 12 Toilets 12-1~12-5 Toilet stalls 13-1~13-5 Toilet 14-1~14-5 Toilet door 15 people (human body)

Claims

1. In an imaging device that uses an optical system to image and capture a moving subject on a linear sensor in which pixels are arranged in one dimension, The pixel arrangement direction of the linear sensor is positioned diagonally with respect to the direction of movement of the subject. The subject, including its shape, contains a correlated pattern in a direction perpendicular to the subject's direction of movement. The process involves capturing a subject pattern and outputting a corresponding pixel output pattern from a linear sensor. A linear sensor repeatedly takes images in a constant drive cycle. Extracting areas from a linear sensor where the pixel output pattern shows significant changes corresponding to the brightness and darkness of the subject pattern. In adjacent drive cycles, extract the areas with significant changes corresponding to the same location in the subject pattern, and determine the pixel position shift of the areas with significant changes in the adjacent drive cycles. Extracting regions where the amount of pixel position shift in areas with large changes remains constant over adjacent drive cycles. Within the region where the extracted pixel position shift amount remains constant, select a pixel position shift amount that corresponds to a correlated pattern in a direction perpendicular to the subject's direction of movement and can be used for the subject's movement speed. The system calculates the subject's movement speed and direction based on the selected pixel position shift amount, optical system information, and the angle between the linear sensor's pixel array direction and the subject's movement direction. An optical measurement system characterized by the following.

2. The location where the pixel output pattern corresponding to the brightness and darkness of the subject pattern output from the linear sensor changes significantly is the extreme value position where the magnitude of the pixel output changes significantly. The optical measurement system according to claim 1, characterized by the following:

3. The location where the pixel output pattern corresponding to the brightness and darkness of the subject pattern output from the linear sensor changes significantly is the extreme value position where the size of the pixel output is minimum or maximum. The optical measurement system according to claim 1, characterized by the following:

4. A method for extracting areas with large changes in the pixel output pattern corresponding to the brightness of the subject pattern output from a linear sensor is to create a pixel output pattern using the moving average value of the pixel output corresponding to the position of the subject and the adjacent pixel output, The optical measurement system according to claim 1, characterized by the following:

5. A method for extracting regions in which the amount of shift in the pixel position of a large change is constant over adjacent drive cycles is to create a pixel output pattern using the moving average of the amount of shift in the pixel position with respect to adjacent pixels. The optical measurement system according to claim 1, characterized by the following:

6. A method for extracting areas from a linear sensor where the pixel output pattern corresponding to the brightness of the subject pattern shows a large change is to perform differential processing on the pixel output pattern in adjacent drive cycles of the linear sensor and use this as the pixel output pattern. The optical measurement system according to claim 1, characterized by the following:

7. A method for selecting a pixel position shift amount that corresponds to a correlated pattern in a direction perpendicular to the direction of movement of the subject and can be used for the movement speed of the subject, by determining using a sign that includes zero for a constant pixel position shift amount. The optical measurement system according to claim 1, characterized by the following:

8. A method for selecting a pixel position shift amount that corresponds to a correlated pattern in a direction perpendicular to the direction of movement of the subject and can be used for the movement speed of the subject, which involves using a constant pixel position shift amount to determine the movement speed of the subject, and comparing and determining the shape of the moving body calculated from the speed. The optical measurement system according to claim 1, characterized by the following:

9. When there is no correlated pattern within the subject, including its shape, in a direction perpendicular to the direction of the subject's movement, a correlated pattern is formed by physical or chemical means in a direction perpendicular to the direction of the subject's movement before the moving subject is imaged and captured by the optical system. The optical measurement system according to claim 1, characterized by the following:

Images