Image processing system and image processing method

Abstract

To provide means that allows distance measurement with a less parallax error and high accuracy in a stereo camera in which the relative relationship between cameras is unknown.SOLUTION: An image processing system comprises a plurality of imaging apparatuses, a first chart plate, a second chart plate, and an image processing apparatus. The image processing apparatus comprises: a position and angle estimation unit that estimates the position and the angle of each of the plurality of imaging apparatuses; a collimated optical axis creation unit that collimates the optical axes of the plurality of imaging apparatuses through image processing; a paralleled image creation unit that creates an image in which the images picked up by the plurality of imaging apparatuses are paralleled; a facing processing unit that causes the images picked up by the plurality of imaging apparatuses to face the second chart plate; a distance difference correction unit that corrects the measured value of the distance from each of the plurality of imaging apparatuses to the second chart plate; a parallax calculation unit that calculates the parallax between the images picked up by the plurality of imaging apparatuses; and a relative angle correction value calculation unit that calculates the relative angle error correction value between the plurality of imaging apparatuses.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an image processing system and method used for calibrating an in-vehicle stereo camera. [Background technology]

[0002] In general, the longer the distance between the left and right cameras, i.e. the longer the baseline length, the more advantageous it is for distance measurement in an on-board stereo camera. In addition, when an on-board stereo camera is used as a sensor for autonomous driving, the optical axes of the two cameras must be aligned with high precision. When the two cameras are housed in the same small housing, the rigidity of the housing allows the optical axis direction to be kept in the same direction with high precision, but to achieve a long baseline length, the two cameras must be housed in separate housings, making it difficult to physically align the optical axes. To address this problem, Patent Document 1 uses a method in which the optical axis misalignment of the camera is determined from the image captured by the camera and the camera is physically moved. In Patent Document 2, the optical axis misalignment is detected by capturing images of the same pattern with the camera at intervals that match the baseline length. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2018-031733 A [Patent Document 2] International Publication No. 2019 / 087253 Summary of the Invention [Problem to be solved by the invention]

[0004] However, providing an on-board stereo camera with a power mechanism for moving the cameras as shown in Patent Document 1 leads to an increase in cost. In addition, the method shown in Patent Document 2 requires two charts to be placed at the same distance as the baseline length, which means that it cannot be used when the exact baseline length is unknown, and the two cameras must be almost directly facing the chart. Thus, there is a demand for technology that can correct the optical axis misalignment for on-board stereo cameras with unknown baseline lengths to achieve highly accurate distance measurements.

[0005] The present invention has been made in consideration of the above problems, and aims to provide a means for enabling highly accurate distance measurement with little parallax error in a stereo camera in which the relative relationship between the cameras is unknown. [Means for solving the problem]

[0006] In order to solve the above-mentioned problems, the present invention provides an image processing system comprising a plurality of image capture devices, a first screen having a repeating pattern drawn thereon and positioned a first distance away from the plurality of image capture devices, a second screen having a repeating pattern and a non-repeating pattern drawn thereon and positioned a second distance away from the plurality of image capture devices, and an image processing device, in which the image processing device includes a position and angle estimating section which estimates the position and angle of each of the plurality of image capture devices based on the results of imaging of the repeating pattern captured by the plurality of image capture devices, a parallelized optical axis creating section which uses the results of estimation by the position and angle estimating section to parallelize the optical axes of the plurality of image capture devices by image processing, and a parallelized optical axis creating section which performs image processing based on the results of processing by the parallelized optical axis creating section. the image capturing apparatus includes a parallelized image creating unit that creates parallelized images of images captured by the multiple image capturing devices; a normalization processing unit that uses the images created by the parallelized image creating unit to normalize the images captured by the multiple image capturing devices with respect to the second screen; a distance difference correction unit that corrects a measured value of a distance from each of the multiple image capturing devices to the second screen by adjusting a magnification of the normalized image using an estimation result from the position and angle estimator; a parallax calculation unit that calculates a parallax of the images captured by the multiple image capturing devices based on an imaging result of a non-repeating pattern that reflects the processing result from the distance difference correction unit; and a relative angle correction value calculation unit that calculates a correction value for a relative angle error between the multiple image capturing devices based on the calculation result from the parallax calculation unit and the second distance. Effect of the Invention

[0007] By using the present invention, it is possible to provide a means for enabling highly accurate distance measurement with little parallax error in a stereo camera in which the relative relationship between the cameras is unknown. Further features related to the present invention will become apparent from the description of the present specification and the accompanying drawings. Furthermore, the problems, configurations and effects other than those described above will become apparent from the following description of the embodiments. [Brief description of the drawings]

[0008] [Figure 1] FIG. 1 is a block diagram showing a functional configuration of an image processing system according to an embodiment of the present invention. [Diagram 2] 4 is a flowchart showing a process executed by an image processing system according to an embodiment of the present invention. [Diagram 3] FIG. 2 shows a screen arrangement according to one embodiment of the present invention. [Figure 4] 1 is an example of a chart arrangement used in the present invention. [Diagram 5] 13 is another example of a chart arrangement used in the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing the overall functional configuration of an image processing system according to an embodiment of the present invention. The image processing system includes vehicle-mounted cameras 101 and 102, an image processing device 100, and multiple screens. The multiple screens specifically refer to first screens 302 and 303, which are arranged at a first distance from camera units 304 and 305 and on which a repeating pattern is drawn, and a second screen 301, which is arranged at a second distance from camera units 304 and 305 that is farther than the first distance and on which a repeating pattern and a non-repeating pattern are drawn, as shown in FIG. 3. The image processing device 100 is composed of a computer equipped with a CPU (Central Processing Unit) and a memory, and has the following functional units:

[0010] That is, the image processing device 100 includes distortion removal units 103 and 104 that correct lens distortion of the wide-angle camera, repeating pattern detection units 105 and 106 that detect a repeating pattern used for calibration from the image after distortion has been removed, camera position angle estimation units 107 and 108 (corresponding to the position angle estimation unit in the claims) that estimate the angle and position of the camera from the repeating pattern, a relative parameter calculation unit 109 that calculates the relative position and relative angle of the two cameras, a parallelization optical axis creation unit 110 that determines the parallelization optical axis from the relative position and relative angle and creates a rotation matrix for parallelization, and a parallelization unit 120 that performs parallelization on the distortion-removed image. the two cameras to the screen; distance difference correction units 115 and 116 which adjust the distance from the two cameras to the screen; a parallax calculation unit 117 which calculates the parallax of the non-repeating pattern from the two aligned images; a relative angle correction value calculation unit 118 which calculates a correction value for the relative angle error between the imaging devices from the parallax; a correction value reflection unit 119 which corrects the relative angle using the correction value; and an output terminal 120 which outputs the corrected relative parameters as calibration values.

[0011] Next, the operation of each part will be described. An in-vehicle stereo camera has two or more camera units, but in this embodiment, two cameras will be described. The two cameras are arranged at different positions, usually on the left and right. In addition, the distortion removal unit 103 to the output terminal 120, which are functional units that input the imaging results of the cameras and output the calculation results, may be arranged independently as separate devices, or may be arranged as a package in which some or all of them are contained in one housing. That is, the image processing system according to this embodiment constitutes one device as a whole in this embodiment, but may have a configuration in which multiple devices are combined. In addition, in order to achieve a long baseline length, the in-vehicle stereo camera may be mounted on a vehicle with two cameras housed in separate housings, and is referred to as a "separate type camera" in this embodiment.

[0012] As described later, images containing repeating and non-repeating patterns are captured by the two vehicle-mounted cameras 101 and 102, but the images obtained are output as images containing distortion caused by the wide-angle lens. Since this distortion is specific to the camera unit, it is converted into a distortion-free image by distortion removal units 103 and 104 using a correction table or the like.

[0013] The images from which the distortion has been removed are sent to repeating pattern detection units 105 and 106, which detect known repeating patterns in each image. In this repeating pattern, known markers are arranged repeatedly, and the positions of each marker on the image are detected. As described below, the repeating patterns are arranged at a plurality of distances, and at least one repeating pattern is drawn on the same plane as a non-repeating pattern.

[0014] The images in which the repeated patterns arranged at multiple distances are detected are sent to camera position angle estimation units 107 and 108. Since the positions of the markers in the image space are known, if there are three or more markers, the three-dimensional position and three angles of the cameras can be calculated by the camera position angle estimation units 107 and 108. The calculation method can be, for example, a method in which a designed position and angle are given as initial values, the projected position on the imaging surface is calculated using these parameters, and the difference from the actual marker detection position is used as an error function, and the parameters are corrected in the direction in which the error function decreases to converge, but the method is not limited to this.

[0015] The positions and angles of the two cameras estimated by the camera position angle estimation units 107 and 108 are sent to a relative parameter calculation unit 109. The relative parameter calculation unit 109 calculates the difference between the positions and angles of the two cameras and outputs it to a parallelized optical axis creation unit 110 as a relative parameter.

[0016] The optical axes of the two cameras 101 and 102 are physically oriented in different directions, and the optical axis directions must be aligned to correctly calculate parallax. High-precision orientation control is required to physically change the orientation of the cameras, which is not realistic for an in-vehicle stereo camera, so a method is used in which images with the orientations of the left and right cameras aligned are created by image processing without changing the physical orientation of the cameras. This processing is called parallelization processing, and the parallelized optical axis creation unit 110 creates the orientation of the virtual optical axes after parallelization processing.

[0017] The direction of the parallelization optical axis can be determined by using one of the left and right cameras as a reference camera and aligning the other camera, or by taking the middle point between the directions of the two optical axes. After parallelization processing, a rotation matrix is ​​created from the angle difference between the optical axis of the original camera and the virtual parallelization optical axis. This rotation matrix is ​​sent to parallelization image creation units 111 and 112. The parallelization image creation units 111 and 112 apply the rotation matrix sent from the parallelization optical axis creation unit 110 to the distortion-free images output from the distortion removal units 103 and 104 to create images aligned with the direction of the virtual parallelization optical axis. The created images are sent to the orientation correction units 113 and 114.

[0018] The orientation correction processors 113 and 114 create a rotation matrix based on the difference between the virtual parallel optical axis and the orientation of the screen, and convert the images according to the rotation matrix to convert the images captured by the two cameras into images that are directly facing the screen and have the same orientation for the left and right cameras. At this time, the images captured by the two cameras are directly facing the screen on which a non-repeating pattern and a repeating pattern are drawn on the same plane. In particular, in the case of a separate camera, unlike a case in which two cameras are housed in one housing, it is difficult to physically ensure the relative relationship between the two cameras, and therefore a deviation in the relative relationship (relative angle, etc.) between the two cameras may occur. The orientation correction processors 113 and 114 perform the orientation process to orient the two cameras to the screen on which the non-repeating pattern is drawn, even if the two cameras are not directly facing the screen, and this allows the parallax calculation unit 117 to calculate the parallax with high accuracy in the subsequent stage. This in turn allows the relative angle correction value calculation unit 118 to calculate the relative angle correction value with high accuracy. Although the rectification process and the rectification process are described here as separate processes, since both are rotation matrix processes, it is possible to create a rotation matrix that combines both processes and perform image transformation at once.

[0019] Since the distances from the two cameras to the screen are different for images that have been subjected to the orientation correction process, the images must be enlarged or reduced. This process is carried out by the distance difference correction units 115 and 116. Specifically, the distance difference correction units 115 and 116 match the magnification of one of the orientation corrected images to that of the other image, thereby making both images equal in size, which is essentially the same process as adjusting the distance between the two cameras and the screen.

[0020] The image after the distance difference correction process is sent to the parallax calculation unit 117. The parallax calculation unit 117 detects parallax using the non-repeating pattern. Here, parallax is an index that indicates the difference in how the imaged object looks from the left and right cameras, and methods of calculating parallax include, for example, the SGM (Semi Global Matching) method and the SAD (Sum of Absolute Difference) method. These calculation methods are already known, so a detailed explanation will be omitted. Since the non-repeating pattern is a known pattern, the known pattern is found from the images obtained from the left and right cameras by correlation processing or the like, and the parallax is calculated by finding the difference in their positions. In principle, the theoretical value of this parallax is determined by the distance from the camera to the screen board with the non-repeating pattern, but the actually measured parallax value has a difference due to an angle estimation error. The relative angle of the two cameras can be corrected by converting this parallax difference into an angle. This calculation is performed by the relative angle correction value calculation unit 118.

[0021] This correction value is sent to correction value reflection unit 119, and a relative parameter reflecting the correction value calculated by relative angle correction value calculation unit 118 is output as the final output from output terminal 120. The angle reflected as the correction value is usually a small angle, but if this correction value is large, it is considered that the reliability of the value calculated by relative parameter calculation unit 109 is low, so if the correction value exceeds a predetermined threshold, it is possible to output an error without outputting the corrected value. In this case, it is possible to enter into abnormality processing, such as starting over from image acquisition or using it as a trigger for a process such as reattaching the camera.

[0022] The value corrected by the correction value reflecting unit 119 can be output directly from the output terminal 120, but it is also possible to verify the corrected value again using a non-repeated pattern. The operation when performing repeated verification will be explained according to the process flow shown in Figure 2. Note that in this flow, two cameras are assumed to be arranged in a left-right direction facing the screen, which is the subject of the image capture.

[0023] First, it is assumed that images captured by both the left and right cameras are input to the distortion removal unit 103 to generate an image with distortion removed. Then, step 201 is an initial estimation process for the camera position and camera angle using a repeating pattern drawn on a chart placed at multiple distances, which will be described later. This corresponds to the processing in the repeating pattern detection units 105 and 106, the camera position angle estimation units 107 and 108, and the relative parameter calculation unit 109. After the processing in step 201 is executed, the process proceeds to step 202.

[0024] In step 202, the left and right camera images are parallelized based on the results obtained in step 201. This corresponds to the processing performed by the parallelization optical axis creation unit 110 and parallelized image creation units 111 and 112. In step 203, which follows, a straightening process is performed on the screen on which both repeating and non-repeating patterns are drawn, which will be described later. This corresponds to the processing performed by the straightening units 113 and 114.

[0025] In step 204, the distance from the camera position determined in step 201 to a screen on which both repeating and non-repeating patterns, described later, are drawn is calculated, and image conversion is performed to eliminate the distance difference between the left and right cameras. This corresponds to the processing of distance difference correction units 115 and 116. In the following processing of step 205, parallax calculation unit 117 uses the image that has been corrected for the distance difference to perform parallax calculation using the non-repeating patterns that have been subjected to the orientation correction processing in step 203.

[0026] Since the theoretical value of parallax is determined by the distance from the camera to the non-repeating pattern that has been subjected to the orientation correction process in step 203, the theoretical value is calculated from the camera position found in step 201. This theoretical value and the parallax value actually obtained from the image will match if the camera position and angle found in step 201 are correct, but if an error exists, a difference will occur accordingly. This difference becomes the correction value for the relative angle, and high accuracy can be obtained because there are only two parameters to be corrected, the camera position and angle. The calculation of this correction value is executed by relative angle correction value calculation unit 118.

[0027] In step 206, the difference is judged, and if the difference is less than a predetermined value, it is determined that the camera position and angle found in step 201 are correct. In the next step, 207, the correction value reflection unit 119 reflects the relative angle equivalent to the difference in the angle found in step 201, and the process ends. This reflected value becomes the final output.

[0028] If the difference is equal to or greater than the specified value in step 206, the correction value reflecting unit 119 counts the number of repetitions up to that point in step 208. If the number of repetitions is equal to or greater than a predetermined specified value, it is determined that the error that has occurred up to that point has become so large that it is no longer possible to correct it, and the process ends as a failure. If the number of repetitions is less than the specified value, the process proceeds to step 209, where the correction value reflecting unit 119 reflects the difference calculated in step 205 in the camera angle. The process from step 202 to step 205 is performed again for this new camera angle, and the difference between the theoretical value and the parallax value obtained from the image is calculated in step 206. If the difference is again equal to or greater than the specified value, the difference is further reflected and the process from step 202 to step 205 is repeated. If the difference does not fall below the specified value even after a certain number of repetitions, the process ends as a failure as determined in step 208.

[0029] According to the flow explained above, the camera position angle is initially estimated using repeating patterns arranged at multiple distances, and then the captured image is aligned with a panel on which both repeating and non-repeating patterns are drawn to correct the distance difference between the left and right cameras, thereby correcting the relative angle error between the cameras. This makes it possible to perform accurate calibration by performing highly accurate parallax correction, even in an in-vehicle stereo camera system in which the relative relationship between the left and right cameras is unknown, for example, when two cameras, such as a separate camera, are mounted on a vehicle in separate housings.

[0030] An example of the screen layout used in one embodiment of the present invention is shown in Figure 3. Figure 3 is a diagram of the layout seen from above. In this embodiment, as mentioned above, 301 is the second screen on which a repeating pattern and a non-repeating pattern are arranged, 302 and 303 are the first screens on which only a repeating pattern is arranged, and 304 and 305 are camera units.

[0031] Examples of repeating and non-repeating patterns on the second screen 301 are shown in Fig. 4 and Fig. 5. In Fig. 4, 401 and 402 are repeating patterns, and 403 is a non-repeating pattern. In Fig. 5, 501 and 502 are non-repeating patterns, and 503 is a repeating pattern. Since the non-repeating pattern needs to be on the same plane as one of the repeating patterns, one of the charts arranged in Fig. 3 is a chart with repeating and non-repeating patterns. By drawing the non-repeating and repeating patterns on the same plane on the second screen 301, it is possible to perform a straightening process on the screen with the repeating pattern and a straightening process on the non-repeating pattern at the same time, which improves the accuracy of the parallax calculation using the non-repeating pattern by the parallax calculation unit 117. As a result, the relative angle error between the cameras can be calculated with high accuracy. Furthermore, in this embodiment, the second screen 301 with the non-repeating pattern is placed farther from the camera than the first screens 302 and 303 with only the repeating patterns. This is because the non-repeating pattern is used for the camera's parallax measurement, and the greater the distance between the camera and the non-repeating pattern, the more accurate the parallax measurement.

[0032] In Fig. 4, the repeating pattern is a lattice pattern of circles, and in Fig. 5, a chessboard-type lattice pattern. However, the pattern is not limited to these examples as long as the characteristic points are easy to detect and the pattern is arranged regularly to prevent detection errors on the image. In addition, the non-repeating pattern is non-repeating in both Fig. 4 and Fig. 5, but there is no problem as long as the same pattern is unlikely to occur in multiple places. Fig. 4 shows an example with two repeating patterns and one non-repeating pattern, and Fig. 5 shows an example with one repeating pattern and two non-repeating patterns, but the number of patterns is not limited to these. Furthermore, in Fig. 3, the chart boards 302 and 303 are composed only of repeating patterns, so they are the same as the repeating patterns shown in Figs. 4 and 5. If the design of the repeating pattern is the same for all charts, the same detection method can be used, so it is preferable to make it the same, but it is also possible to use different designs.

[0033] In Figure 3, all the screens are set up vertically on the floor so that the camera faces them almost directly, but the first screen (302 and 303 in Figure 3), which is made up only of a repeating pattern, does not need to be set up vertically as it does not need to be directly facing the camera. For example, it is also possible to place the screen on the floor or ceiling. If there is not enough space horizontally, it is also possible to attach it to the wall.

[0034] The above-described embodiment of the present invention provides the following advantageous effects.

[0035] (1) The image processing system according to the present invention comprises a plurality of image capturing devices, a first screen having a first repeating pattern drawn thereon and positioned at a first distance from the plurality of image capturing devices, a second screen having a repeating pattern and a non-repeating pattern drawn thereon and positioned at a second distance from the plurality of image capturing devices, and an image processing device, in which the image processing device comprises a position and angle estimation unit which estimates the position and angle of each of the plurality of image capturing devices based on the imaging results of the repeating pattern captured by the plurality of image capturing devices, a parallelized optical axis creation unit which parallelizes the optical axes of the plurality of image capturing devices by image processing using the estimation results of the position and angle estimation unit, and a parallelized optical axis creation unit which performs image processing based on the processing results of the parallelized optical axis creation unit. the image capturing apparatus includes a parallelized image creating unit that creates parallelized images of images captured by the multiple image capturing devices; a normalization processing unit that uses the images created by the parallelized image creating unit to normalize the images captured by the multiple image capturing devices with respect to the second screen; a distance difference correction unit that corrects a measured value of a distance from each of the multiple image capturing devices to the second screen by adjusting a magnification of the normalized image using an estimation result from the position and angle estimator; a parallax calculation unit that calculates a parallax of the images captured by the multiple image capturing devices based on an imaging result of a non-repeating pattern that reflects the processing result from the distance difference correction unit; and a relative angle correction value calculation unit that calculates a correction value for a relative angle error between the multiple image capturing devices based on the calculation result from the parallax calculation unit and the second distance.

[0036] The above configuration makes it possible to provide a means for enabling highly accurate distance measurement with little parallax error in a stereo camera in which the relative relationship between the cameras is unknown.

[0037] (2) The second distance is greater than the first distance. By placing the second chart at a distance greater than the first distance, the parallax between the imaging devices can be measured more accurately, and as a result, the relative error between the imaging devices can be corrected more accurately.

[0038] (3) If the relative angle error correction value calculated by the relative angle correction value calculation unit is greater than a predetermined threshold, the image processing device issues an error, making it possible to decide whether to discard the correction value and start over from image capture.

[0039] (4) When the relative angle error correction value calculated by the relative angle correction value calculation unit is greater than a predetermined threshold, the image processing device reflects the relative angle error correction value in the multiple image capture devices and then executes image processing again, thereby making it possible to repeatedly fine-tune the image capture devices and improve the accuracy of the correction.

[0040] (5) The image processing method according to the present invention estimates the position and angle of each of the imaging devices based on the results of imaging a repeating pattern on a first screen and a second screen located a first distance and a second distance away from the imaging devices, collimates the optical axes of the imaging devices using the estimation results, orients the images captured by the imaging devices relative to the second screen using the collimation results, corrects the measured distance from each of the imaging devices to the second screen by adjusting the magnification of the orientated images using the estimation results, calculates the parallax of the images captured by the imaging devices based on the results of imaging a non-repeating pattern on the second screen with the distance measurements corrected, and calculates a correction value for the relative angle error between the imaging devices based on the calculated parallax and the second distance. This provides the same effects as (1). The method according to the present invention also has the same configurations and effects as (2) to (4).

[0041] The present invention is not limited to the above-described embodiments, and various modifications are possible. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and the present invention is not necessarily limited to an embodiment having all of the configurations described. It is also possible to replace a part of the configuration of one embodiment with the configuration of another embodiment. It is also possible to add the configuration of another embodiment to the configuration of one embodiment. It is also possible to delete a part of the configuration of each embodiment, or to add or replace another configuration. [Explanation of symbols]

[0042] 101, 102 camera (imaging device), 107, 108 camera position angle estimating unit (position angle estimating unit), 110 parallelized optical axis creating unit, 111, 112 parallelized image creating unit, 113, 114 orientation processing unit, 115, 116 distance difference correcting unit, 117 parallax calculating unit, 118 relative angle correction value calculating unit, 301 second screen, 302, 303 first screen

Claims

1. A plurality of imaging devices; a first screen having a repeating pattern and positioned a first distance from the plurality of image capture devices; and a second screen having a repeating pattern and a non-repeating pattern and positioned a second distance from the plurality of image capture devices. An image processing system comprising: The image processing device includes: a position and angle estimation unit that estimates a position and an angle of each of the plurality of imaging devices based on an imaging result of the repeating pattern captured by the plurality of imaging devices; a parallelized optical axis creation unit that parallelizes the optical axes of the plurality of image capture devices by image processing using the estimation result of the position angle estimation unit; a parallelized image creation unit that creates a parallelized image by parallelizing the images captured by the plurality of image capture devices based on a processing result of the parallelization optical axis creation unit; a rectification unit which rectifies the images captured by the plurality of image capture devices relative to the second screen using the images created by the rectification image creation unit; a distance difference correction section which corrects the measured distance from each of the plurality of image capture devices to the second screen by adjusting the magnification of the orthogonalized image using the result of estimation by the position and angle estimator; a parallax calculation unit that calculates parallax of images captured by the plurality of imaging devices based on an imaging result of the non-repeated pattern reflecting a processing result of the distance difference correction unit; a relative angle correction value calculation unit that calculates a relative angle error correction value between the plurality of image capture devices based on a calculation result of the parallax calculation unit and the second distance.

1. An image processing system comprising:

2. 2. The image processing system according to claim 1, the second distance is greater than the first distance; 1. An image processing system comprising:

3. 2. The image processing system according to claim 1, The image processing device issues an error when the relative angle error correction value calculated by the relative angle correction value calculation unit is greater than a predetermined threshold value.

1. An image processing system comprising:

4. 2. The image processing system according to claim 1, when the relative angle error correction value calculated by the relative angle correction value calculation unit is greater than a predetermined threshold value, the image processing device reflects the relative angle error correction value in the plurality of image capturing devices and then performs image processing again.

1. An image processing system comprising:

5. 1. An image processing method performed by an image processing system, comprising: estimating the position and angle of each of the plurality of image capturing devices based on the captured images of a repeating pattern depicted on a first screen and a second screen that are positioned at a first distance and a second distance from the plurality of image capturing devices; Using the result of the estimation, the optical axes of the plurality of imaging devices are collimated; Using the result of the parallelization, the images captured by the plurality of image capturing devices are aligned in a normal direction relative to the second screen; using the result of said estimation to correct a measurement of the distance from each of said plurality of image capture devices to said second screen by adjusting a magnification of said orthogonalized image; calculating a parallax of the images captured by the plurality of image capture devices based on the image capture results of the non-repeating pattern drawn on the second screen after the distance measurement values ​​have been corrected; calculating a relative angle error correction value between the plurality of image capturing devices based on the calculated parallax and the second distance; 13. An image processing method comprising:

6. 6. The image processing method according to claim 5, The second distance is greater than the first distance.

13. An image processing method comprising:

7. 6. The image processing method according to claim 5, if the calculated relative angle error correction value is greater than a predetermined threshold, issue an error.

13. An image processing method comprising:

8. 6. The image processing method according to claim 5, if the calculated relative angle error correction value is greater than a predetermined threshold, the relative angle error correction value is reflected in the plurality of image capture devices, and then image processing is performed again.

13. An image processing method comprising:

Images