Multi-camera testing method, system and storage medium

Abstract

The present invention relates to the field of camera testing technology, specifically a multi-camera testing method, system and storage medium, the method comprising: S1: obtaining the original Raw image collected by the module to be tested during the test process; S2: performing image recognition based on the original Raw image to obtain the coordinates of the four corner points of the grid where the geometric center of each camera is located, and calculating the geometric center of the original Raw image; S3: assigning the corresponding two-dimensional coordinates of the grid in units of grids, and calculating the two-dimensional coordinates of the grid where the geometric center of each camera is located; S4: taking any camera as a reference, calculating the relative position of the geometric center of other cameras and the geometric center of the reference camera; S5: calculating the offset angle of the geometric center of each camera based on the relative position of the geometric center, and judging whether the offset angle is greater than the offset angle threshold. If so, the test fails, otherwise, the test passes. By adopting this solution, effective detection of multi-cameras can be achieved.

Description

Technical Field

[0001] The present invention relates to the field of camera testing technology, and in particular to a multi-camera testing method, system and storage medium. Background Art

[0002] Currently, cameras are widely used in various fields such as mobile phones, vehicles, medical treatment, security, AIOT (artificial intelligence Internet of Things), etc. However, monocular cameras are often unable to meet actual needs, so multi-cameras, such as binocular cameras, have gradually emerged. With the rapid development of binocular camera technology, different binocular camera configurations can be used to calculate depth of field, achieve background blur and refocus, improve low-light image shooting quality, optical zoom, three-dimensional reconstruction and other functions. For camera module manufacturers, it is particularly important to ensure the optical axis offset angle between multi-cameras. If the optical axis offset angle between multi-cameras is too large, it will lead to poor dual-camera fusion effect, image blur and other problems. Therefore, there is an urgent need for a multi-camera testing method, system and storage medium that can effectively detect multi-cameras. Summary of the Invention

[0003] One of the purposes of the present invention is to provide a multi-camera testing method, system and storage medium that can effectively detect multi-cameras.

[0004] The present invention provides a basic solution 1: a multi-camera testing method, including the following contents:

[0005] S0: During the test, the module under test captures the target module. The module under test includes multiple cameras, and the target module is provided with a checkerboard pattern with alternating black and white grids.

[0006] S1: Obtain the original Raw image collected by the module under test during the test process;

[0007] S2: Perform image recognition based on the original raw image to obtain the coordinates of the four corner points of the grid where the geometric center of each camera is located, and calculate the geometric center of the original raw image;

[0008] S3: Taking the grid as the unit, assign the corresponding two-dimensional coordinates of the grid in turn, and calculate the two-dimensional coordinates of the grid where the geometric center of each camera is located;

[0009] S4: Taking any camera as a reference, calculate the relative position of the geometric center of other cameras and the geometric center of the reference camera;

[0010] S5: Calculate the offset angle of the geometric centers of the two cameras based on the relative positions of the geometric centers, and determine whether the offset angle is greater than the offset angle threshold. If so, the test fails; otherwise, the test passes.

[0011] Beneficial effects of basic plan 1:

[0012] Take binocular camera as an example, as shown in the attached Figure 1 As shown in the figure, the geometric offset angle of the binocular camera: the deviation of the geometric centers of the CMOS sensors of the two cameras. The optical axis offset angle of the binocular camera: the lens and the CMOS sensor are not completely parallel and may not fall at the geometric center (yellow dot). Therefore, the geometric offset angle of the binocular camera is used to align the optical axis offset angle of the binocular camera.

[0013] This solution uses image detection to identify the two-dimensional spatial location of the geometric center, accurately calculating the geometric offset angle of the multi-camera system and calibrating the optical axis offset angle. Based on the calculated offset angle, a test is performed. If the offset angle exceeds a threshold, the multi-camera offset angle is too large and the test fails. Otherwise, the test passes. This solution enables efficient and flexible determination of multi-camera offset angles, enabling effective multi-camera testing.

[0014] Furthermore, S2 includes the following:

[0015] The original raw image is cropped, and the cropped image contains the grid where the geometric center is located.

[0016] Beneficial effects: By cropping the image, the number of irrelevant images that need to be processed is reduced, and image recognition and detection processing is performed only on the middle part of the image, which reduces the test time and improves UPH (productivity per unit hour).

[0017] Furthermore, S2 also includes the following:

[0018] The cropped image is filtered, grayscale converted, corner detected, and sub-pixel corner detected in sequence;

[0019] Count the number of corner points and determine whether the number of corner points is lower than the preset number.

[0020] Beneficial effects: The cropped image is filtered to reduce interference. At the same time, corner point statistics are used to determine whether the cropping is incorrect or the collected image is incorrect. For example, the preset number is 4. When the number of corner points is less than 4, the cropped image does not contain a complete grid and does not meet the image requirements.

[0021] Furthermore, S2 also includes the following:

[0022] If the number of corner points is not less than the preset number, calculate the distance of each corner point from the geometric center and sort the corner points from low to high based on the distance;

[0023] Classify the coordinates of the first four corner points and determine them as the upper left corner, upper right corner, lower left corner, and lower right corner of the geometric center of the cropped image in turn;

[0024] The average side length in the horizontal direction and the average side length in the vertical direction are calculated based on the coordinates of the first four corners, and it is determined whether the difference between the two average side lengths is within a preset range.

[0025] Beneficial effect: The coordinates of the four corner points of the grid where the geometric center is located are obtained, and the average side length of each grid is calculated to facilitate subsequent offset angle calculation.

[0026] Furthermore, S3 includes the following:

[0027] The center coordinates of the grid where the geometric center of each camera is located are calculated based on the coordinates of the first four corner points, and the extension point coordinates are generated by expansion based on the center coordinates.

[0028] Beneficial effect: Taking into account the individual differences in AA active focusing of multi-cameras, the geometric center of the entire image may be on the edge of the black or white grid. In addition, there are differences in the placement of multi-cameras on the multi-camera control machine. Therefore, the center coordinates of the grid are used to offset it up, down, left, and right by a certain value to obtain the pixel values ​​of the grid center extension point and its nearby points to ensure that it is not affected by the surrounding environment.

[0029] Furthermore, S4 includes the following:

[0030] Calculate the horizontal and vertical scale coefficients of other cameras and the reference camera;

[0031] The relative positions of the geometric centers of other cameras and the reference camera are calculated based on the four corner point positions, two-dimensional coordinates, and horizontal and vertical axis scale factors to generate geometric center relative pixel coordinates (x_diff, y_diff).

[0032] Beneficial effect: The horizontal and vertical scale coefficients of other cameras and the reference camera are calculated, for example, the horizontal and vertical scale coefficients of the first camera with the second camera as a reference, so as to generate the relative pixel coordinates of the geometric center.

[0033] Furthermore, S5 includes the following:

[0034] The offset angle of the geometric center is calculated according to the following formula:

[0035]

[0036] Where PixelSize is the pixel size of the chip sensor, EFL is the effective focal length of the lens, and π is the pi.

[0037] Beneficial effect: The offset angle of the geometric center is calculated using this formula.

[0038] A second object of the present invention is to provide a multi-camera testing system.

[0039] The present invention provides a second basic solution: a multi-camera testing system, including a target module, a control machine and a server;

[0040] The target module is equipped with a checkerboard pattern with alternating black and white grids. The target module is placed at a preset distance in front of the module to be tested.

[0041] The control machine is equipped with a module to be tested, which is used to obtain and upload image data during the test process. The image data is the original Raw image.

[0042] The server is used to calculate the image data according to S1 to S5 in the above-mentioned multi-camera testing method and output the geometric offset angle.

[0043] Basic Option 2 Beneficial Effects:

[0044] The target module is set up to provide the chart required for the test, control the machine settings, and fix the module to be tested. The server uses a host computer equipped with the corresponding program to analyze the image data collected by the module to be tested, thereby deriving the geometric offset angle of the module to be tested and realizing effective detection of multiple cameras.

[0045] Furthermore, it also includes a light source module, which is used to provide brightness for the target module.

[0046] Beneficial effect: The light source module is placed behind or in front of the target board to ensure that the brightness of the target board is appropriate, and that the image captured by the multi-eye camera can ensure that the checkerboard is clearly visible, while avoiding overexposure or underexposure of the target board.

[0047] A third object of the present invention is to provide a storage medium.

[0048] The present invention provides a third basic solution: a storage medium storing executable computer instructions, which, when executed, uses S1 to S5 in the multi-camera testing method described in any one of claims 1 to 7. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a schematic diagram of the present invention;

[0050] Figure 2 This is a schematic structural diagram of an embodiment of a multi-camera testing system of the present invention;

[0051] Figure 3 A schematic diagram of a chessboard taken during the testing process of the present invention;

[0052] Figure 4 This is a flow chart of an embodiment of a multi-camera testing method of the present invention;

[0053] Figure 5 A schematic diagram of corner points of an embodiment of a multi-camera testing method, system, and storage medium of the present invention;

[0054] Figure 6 This is a schematic diagram of the center point expansion of an embodiment of a multi-camera testing method, system, and storage medium of the present invention. DETAILED DESCRIPTION

[0055] The following is further described in detail through specific implementation methods:

[0056] Example

[0057] A multi-camera testing system, as shown in the attached Figure 2 As shown, it includes a light source module, a target plate module, a control machine and a server.

[0058] As attached Figure 3 As shown, the target module features a checkerboard pattern of alternating black and white squares. A chart consisting of these checkerboard patterns is created, with each square being assigned a corresponding two-dimensional coordinate (y, x) based on the top left corner of the square. (Y represents the yth row and x represents the xth column.) The target module is placed at a predetermined distance directly in front of the module under test.

[0059] The light source module is used to provide brightness for the target module. The light source module is set in front of or behind the target module according to actual conditions to ensure that the brightness of the checkerboard is appropriate and that the image captured by the multi-eye camera can ensure that the checkerboard is clearly visible.

[0060] The control machine is equipped with a module to be tested, which is used to capture and upload raw image data during the test. Specifically, the control machine includes a multi-camera, a fixture equipped with the multi-camera, a test box, cables connecting the fixture to the test box, and a data cable connecting the test box to a computer. The control machine must be adjusted to maintain a fixed distance between the chart and the control machine, ensuring that the fixture equipped with the multi-camera is parallel to the chart board module. Images captured by the multi-camera must include the entire checkerboard chart.

[0061] The server is used to calculate the image data and output the geometric offset angle according to S1 to S5 in the following multi-camera test method. Specifically, the image data of each camera is captured separately. Using a certain camera A as a reference, the geometric center of its image is ensured to be near the chart. The other cameras are then adjusted to ensure that the geometric centers of the images of the other cameras can also fall on the checkerboard chart. If it cannot be guaranteed that the geometric centers of the images of other cameras can also fall on the checkerboard chart, it means that the distance between the chart and the multi-camera control machine is not appropriate or the offset angle of the multi-camera itself is too large. The image data of the multi-camera is collected, and the geometric center of each camera is located on the specific coordinates of the checkerboard through image detection and recognition. Using a certain camera A as a reference, the coordinates of the other cameras relative to the certain camera A are calculated, thereby calculating the geometric offset angles between the multi-cameras.

[0062] A multi-camera testing method is applied to the multi-camera testing system mentioned above, as shown in the attached Figure 4 As shown, including the following:

[0063] S0: During the test, the module under test captures the target module. The module under test includes multiple cameras, and the target module is provided with a checkerboard pattern with alternating black and white grids.

[0064] S1: Get the original Raw image Rawbuf collected by the module under test during the test process i [w i *h i ];

[0065] S2: Based on the original Raw image, image recognition is performed to obtain the coordinates of the four corner points of the grid where the geometric center of each camera is located And calculate the geometric center of the original Raw image

[0066] S3: Taking the grid as the unit, assign the corresponding two-dimensional coordinates of the grid in turn, and calculate the two-dimensional coordinates (X i ,Y i );

[0067] S4: Taking any camera as a reference, calculate the relative position of the geometric center of other cameras and the geometric center of the reference camera;

[0068] S5: Calculate the offset angle of the geometric centers of the two cameras based on the relative positions of the geometric centers, and determine whether the offset angle is greater than the offset angle threshold. If so, the test fails; otherwise, the test passes.

[0069] S1 includes the following:

[0070] Capture the clear original Raw image Rawbuf of multiple cameras i [w i *h i ], i is the i-th camera, w i 、h i are pixel width and pixel height respectively.

[0071] S2 includes the following:

[0072] S2-1: Crop the original Raw image. The cropped image contains the grid where the geometric center is located. According to the principle of cropping from the center to the surrounding area, the cropped image is Rawbuf i [(w i -c1 i )*(h i -c2 i )],c1 i 、c2 i The cropping values ​​are pixel width and pixel width respectively. By cropping the image, only the middle part of the image is processed for image recognition and detection, which reduces image processing time.

[0073] S2-2: Filter the cropped image. In this embodiment, Gaussian filtering is performed on the cropped image, and linear smoothing filtering is performed to obtain the cropped Gaussian filter image Rawbuf_Gaussian i [(w i -c1 i )*(h i -c2 i )].

[0074] S2-3: grayscale conversion is performed on the filtered image. In this embodiment, the image Rawbuf_Gaussian_Gray is obtained after grayscale conversion of the cropped Gaussian filter image. i [(w i -c1 i )*(h i -c2 i )], the center coordinates of the grayscale converted image are (center_x i ,center_y i ):

[0075]

[0076] S2-4: As attached Figure 5As shown in the figure, the image after grayscale conversion is subjected to corner detection and sub-pixel corner detection. In this embodiment, the corner detection function goodFeaturesToTrack() in OpenCV is used to set the corner points in the detection image, and the sub-pixel detection function cornerSubPix() in OpenCV is used to obtain the corner coordinates (corner_x im ,corner_y im ), corner_x im 、corner_y im The horizontal and vertical coordinates of the mth corner point are calculated by corner point detection in the image cropped by the i-th camera after Gaussian filtering and grayscale conversion.

[0077] S2-5: Count the number of corner points and determine whether the number of corner points is less than the preset number. If not, calculate the distance of each corner point from the geometric center and sort the corner points from low to high according to the distance. In this embodiment, the preset number is 4. If the number of corner points is less than 4, it means that the number of grids in the cropped image is incorrect and the image should be cropped again. Otherwise, calculate the distance from each corner point to the geometric center and sort them from low to high. Each corner point (corner_x i ' m ,corner_y i ' m ) distance from the geometric center im ;

[0078] distance i0 ≤distance i1 ≤distance i2 ...≤distance im ;

[0079]

[0080] S2-6: Classify the coordinates of the first four corner points (corner_x i ' k ,corner_y i ' k ), k=0,1,2,3, according to the following formula, the coordinates of the corner points are the coordinates of the upper left corner of the geometric center (corner_x0 i ,corner_y0 i ), the coordinates of the upper right corner (corner_x1 i ,corner_y1 i ), the coordinates of the lower left corner (corner_x2 i ,corner_y2i ), the coordinates of the lower right corner (corner_x3 i ,corner_y3 i ):

[0081]

[0082]

[0083]

[0084]

[0085] S2-7: Calculate the mean side length x_side_ave in the horizontal direction based on the first four corner coordinates and the following formula i and the mean side length y_side_ave in the vertical direction i , determine whether the difference between the two mean side lengths is within the preset range. If not, it means that the first four corners do not form a square, and recapture the original Raw image.

[0086]

[0087] S2-8: Restore the coordinates of the center points of the four corners in the original raw image based on the coordinates of the first four corners Original Raw image Rawbuf i [w i *h i The geometric center of ] is where represents the actual coordinates of the i-th camera k=0 upper left corner, k=1 upper right corner, k=2 lower left corner, and k=3 lower right corner in the original Raw image.

[0088] S3 includes the following:

[0089] S3-1: The geometric center of each camera falls in the grid determined by the first four corner points, and the center coordinates of this grid are calculated based on the coordinates of the first four corner points. and the mean side length x_side_ave in the horizontal direction i , the mean side length y_side_ave in the vertical direction i , calculate the center coordinates of the grid where the geometric center of each camera is located as:

[0090]

[0091] S3-2: Considering individual differences in AA active focus adjustment among multiple cameras, the geometric center of the entire image may fall on the edge of a black or white grid. Furthermore, due to the variability in the placement of multiple cameras on the control panel, the center coordinates of the grid are offset upward, downward, and leftward by a certain value to determine the pixel values ​​of the grid center extension point and its surrounding points. This ensures that the grid is not affected by the surrounding environment. The mean side lengths along the horizontal and vertical axes are used as unit steps. The pixel feature values ​​of the grid center point are used to determine whether it falls on a black or white grid. The feature values ​​decrease gradually toward the left and above the center of the grid. Classification is used to determine the two-dimensional coordinates of the grid on the checkerboard chart. Furthermore, a checkerboard boundary is determined by determining whether the color of the geometric center grid is inconsistent with that of the grids to its left or right. The color of the grid at the geometric center of the image is determined using the pixel values ​​of several locations at the center extension point of the grid. Taking the black grid as the grid where the geometric center of the image is located as an example, each grid has a center cross, S0\S1\S2\S3 are the grid center extension points, and C is the grid center, as shown in the attached figure. Figure 6 shown.

[0092] Coordinates of the four extension points S0\S1\S2\S3:

[0093]

[0094] In the formula, according to empirical values, n = 2, 3, 4, 5.

[0095] S3-3: Get the horizontal coordinate X and vertical coordinate Y of the geometric center of the image: Taking the black grid as the grid where the geometric center of the image is located as an example, first determine whether the extension point of the black grid and the points near the extension point are both less than the threshold value, and whether the upper or lower, left or right adjacent grids (that is, the upper left or upper right or lower left or lower right) are all greater than their threshold values, so as to determine that the grid where the geometric center of the image is located is black.

[0096] The eight coordinates of the left and right adjacent grids of the grid where the geometric center of the image is located are:

[0097]

[0098] The eight coordinates of the upper and lower adjacent grids of the grid where the geometric center of the image is located are:

[0099]

[0100] Where m1 = 0, 1, ..., X; m2 = 0, 1, ..., Y.

[0101] Get the vertical coordinate of the geometric center of the image as 0, and the mean side length x_side_ave in the horizontal direction i and the mean side length y_side_ave in the vertical directioni The unit step lengths are respectively, and the left is judged to see whether it is a white grid. The judgment basis is whether the white grid extension point and the points near the extension point are both greater than the threshold value, and whether the upper or lower, left or right adjacent grids (that is, the upper left or upper right or lower left or lower right) are all less than the threshold. In this way, it is determined that the grid where the geometric center of the image is white, and the vertical coordinate is cumulatively added by 1 in sequence until this relationship no longer exists, and the final vertical coordinate is obtained as X.

[0102] The horizontal coordinate of the geometric center of the image is 0, and the mean side length x_side_ave in the horizontal direction is also obtained. i and the mean side length y_side_ave in the vertical direction i The unit step size is used to determine whether it is a white grid. The judgment basis is whether the white grid extension point and the points near the extension point are greater than the threshold value at the same time, and whether the upper or lower, left or right adjacent grids (that is, the upper left or upper right or the lower left or lower right) are all less than the threshold. In this way, it is determined that the grid where the geometric center of the image is located is white, and the vertical coordinate is accumulated by 1 in sequence until this relationship no longer exists, and the final vertical coordinate is obtained as Y.

[0103] S4 includes the following:

[0104] S4-1: The horizontal and vertical axes of the pixels are scaled up, and the horizontal and vertical scale coefficients α and β of the i=1 camera with the i=0 camera as a reference are calculated respectively:

[0105]

[0106]

[0107] S4-2: According to the four corner points of the black or white square grid where the geometric center of each camera is located, the two-dimensional coordinates (X i ,Y i ), and the horizontal and vertical scale coefficients of the i=1th camera with the i=0th camera as a reference are used to calculate the relative pixel coordinates (x_diff, y_diff) of the geometric centers of the i=1th camera and the i=0th camera:

[0108]

[0109]

[0110] S5 includes the following:

[0111] The offset angle of the geometric center of each camera is calculated based on the relative pixel coordinates of the geometric center of each camera and the offset angle formula. It is determined whether the offset angle is greater than the set offset angle threshold. If it is, it means that the offset angle of the multi-camera is too large and the test fails; otherwise, the test succeeds and meets the requirements.

[0112] The offset angle of the geometric center is:

[0113]

[0114] Where PixelSize is the pixel size of the chip sensor with the i=0th camera as a reference, and its unit is um, EFL is the effective focal length of the lens with the i=0th camera as a reference, and π is pi.

[0115] The specific implementation process is as follows:

[0116] Taking the binocular camera project as an example, the main camera sensor IMX586 has a Pixel Size of 0.8μm, an effective focal length (EFL) of 11.8mm, and an image resolution of 7680*4320. The secondary camera sensor IMX700 has a Pixel Size of 1.22μm, an effective focal length (EFL) of 6.67mm, and an image resolution of 7680*4320.

[0117] Using a binocular camera capture as a prototype, capture the clear original raw images of the main camera and the secondary camera respectively, and calculate the coordinates of the four corner points of the grid where the geometric centers of the main camera and the secondary camera are located, as well as the two-dimensional coordinates of the grid where the geometric centers fall.

[0118] The primary and secondary cameras were cropped to the center tenth, and corner detection was performed using OpenCV's goodFeaturesToTrack() method. The maximum number of corner points was set to 200, the quality factor of the corner points was set to 0.1, the minimum Euclidean distance between the returned corner points was set to 20, and the average block size for calculating the derivative covariance matrix for each pixel neighborhood was set to 9. The sub-pixel detection method cornerSubPix() in OpenCV was then used to obtain the coordinates of all corner points. These points were sorted from lowest to highest distance from the image's geometric center, and the coordinates of the top left, top right, bottom left, and bottom right corners of the image's geometric center were obtained.

[0119] The coordinates of the main camera's upper left corner, upper right corner, lower left corner, and lower right corner are (3805.41, 2081.79), (3921.60, 2082.69), (3804.51, 2198.38), and (3920.49, 2199.14), respectively. The center coordinates of the grid where the geometric center of the image is located are (3863, 2139), and the two-dimensional coordinates of the grid are (10, 5). The mean side length along the horizontal axis is calculated to be 116.08, and the mean side length along the vertical axis is calculated to be 116.52.

[0120] The coordinates of the upper left corner, upper right corner, lower left corner, and lower right corner of the secondary camera are (3806.73, 2138.21), (3850.31, 2138.42), (3806.54, 2138.85), and (3849.86, 2181.93), respectively. The center coordinates of the grid where the geometric center of the image is located are (3863, 2139), and the two-dimensional coordinates of the grid are (9, 5). The mean side length in the horizontal direction is 43.45, and the mean side length in the vertical direction is 43.57.

[0121] Using the primary camera as a reference, the secondary camera is mapped onto the primary camera. The horizontal and vertical pixel offsets of the geometric centers of the primary and secondary cameras are calculated to be -61.78 and -19.94, respectively. The offset angle formula is used to determine the geometric center offset to be 0.2522°. The preset threshold is 1°, which meets the requirements.

[0122] A storage medium stores executable computer instructions, which, when executed, use S1 to S5 in the multi-camera testing method.

[0123] The above is only an embodiment of the present invention. Common knowledge such as the known specific structures and characteristics in the scheme is not described in detail here. Ordinary technicians in the field are aware of all common technical knowledge in the technical field of the invention before the application date or priority date, can obtain all existing technologies in the field, and have the ability to apply conventional experimental means before that date. Ordinary technicians in the field can improve and implement this scheme in combination with their own abilities under the inspiration given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention. These will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A multi-camera testing method, characterized by: Also included: S0: During the test, the module under test captures the target module. The module under test includes multiple cameras, and the target module is provided with a checkerboard pattern with alternating black and white grids. S1: Obtain the original Raw image collected by the module under test during the test process; S2: Perform image recognition based on the original raw image to obtain the coordinates of the four corner points of the grid where the geometric center of each camera is located, and calculate the geometric center of the original raw image; S3: Taking the grid as the unit, assign the corresponding two-dimensional coordinates of the grid in turn, and calculate the two-dimensional coordinates of the grid where the geometric center of each camera is located; S4: Taking any camera as a reference, calculate the relative position of the geometric center of other cameras and the geometric center of the reference camera; S5: Calculate the offset angle of the geometric centers of the two cameras based on the relative positions of the geometric centers, and determine whether the offset angle is greater than the offset angle threshold. If so, the test fails; otherwise, the test passes. S4 includes the following: Calculate the horizontal and vertical scale coefficients of other cameras and the reference camera; Calculate the relative position of the geometric center of the other camera and the reference camera based on the four corner point positions, two-dimensional coordinates, and horizontal and vertical axis scale factors to generate the geometric center relative pixel coordinates ; S5 includes the following: The offset angle of the geometric center is calculated according to the following formula: In the formula is the pixel size of the chip sensor, is the effective focal length of the lens, is pi.

2. A multi-camera testing method according to claim 1, characterized in that: S2 includes the following: The original raw image is cropped, and the cropped image contains the grid where the geometric center is located.

3. A multi-camera testing method according to claim 2, characterized in that: S2 also includes the following: The cropped image is filtered, grayscale converted, corner detected, and sub-pixel corner detected in sequence; Count the number of corner points and determine whether the number of corner points is lower than the preset number.

4. A multi-camera testing method according to claim 3, characterized in that: S2 also includes the following: If the number of corner points is not less than the preset number, calculate the distance of each corner point from the geometric center and sort the corner points from low to high based on the distance; Classify the coordinates of the first four corner points and determine them as the upper left corner, upper right corner, lower left corner, and lower right corner of the geometric center of the cropped image in turn; The average side length in the horizontal direction and the average side length in the vertical direction are calculated based on the coordinates of the first four corners, and it is determined whether the difference between the two average side lengths is within a preset range.

5. A multi-camera testing method according to claim 4, characterized in that: S3 includes the following: The center coordinates of the grid where the geometric center of each camera is located are calculated based on the coordinates of the first four corner points, and the extension point coordinates are generated by expansion based on the center coordinates.

6. A multi-camera testing system, characterized by: Including standard board module, control machine and server; The target module is equipped with a checkerboard pattern with alternating black and white grids. The target module is placed at a preset distance in front of the module to be tested. The control machine is equipped with a module to be tested, which is used to obtain and upload image data during the test process. The image data is the original Raw image. The server is used to calculate the image data and output the geometric offset angle according to S1 to S5 in the multi-camera testing method according to any one of claims 1-5.

7. A multi-camera testing system according to claim 6, characterized in that: It also includes a light source module, which is used to provide brightness for the target module.

8. A storage medium, characterized in that: Executable computer instructions are stored, and when executed, S1 to S5 of the multi-camera testing method described in any one of claims 1 to 5 are used.

Images