Monocular camera imaging consistency detection device and method based on relative pixel coordinates

Abstract

The application belongs to the field of monocular camera imaging consistency detection, and particularly relates to a monocular camera imaging consistency detection device and method based on relative pixel coordinates. The device comprises a strapdown fixing tool, a detection target plate and imaging consistency calculation software. The strapdown fixing tool comprises a monocular camera strapdown fixing tool and a detection target plate strapdown fixing tool. The detection target plate has a pattern for camera detection on one side. The pattern is a fixed-length square arranged in a black-and-white chessboard pattern, and the number of rows of the chessboard pattern is less than the number of columns. The detection target plate is installed on the detection target plate strapdown fixing tool. The monocular camera is installed on the monocular camera strapdown fixing tool, connected to an image processing computer through a special cable and subjected to data processing and image consistency calculation through the imaging consistency calculation software. The device has low requirements for detection equipment, is simple to operate, is not easily affected by the environment, is low in cost and convenient to apply.

Description

Device and Method for Imaging Consistency Detection of Monocular Cameras Based on Relative Pixel Coordinates Technical Field

[0001] This invention belongs to the field of monocular camera imaging consistency detection, specifically relating to a monocular camera imaging consistency detection device and method based on relative pixel coordinates. Background Technology

[0002] As an important sensor capable of stably acquiring environmental visual information, monocular cameras are now widely used in fields such as intelligent driving, pattern recognition, and target tracking, playing an increasingly important role in various intelligent systems such as unmanned vehicles and drones. In some application scenarios, monocular cameras need to ensure imaging performance after being subjected to flight vibrations and strong impacts, so imaging consistency detection of monocular cameras has become a necessary step in engineering applications. Currently, the mainstream method for imaging consistency detection of monocular cameras is to use optical instruments for detection, that is, to use collimators to measure the camera's optical path, thereby detecting the principal optical axis offset of the camera before and after vibration and impact, and thus detecting the imaging consistency of the camera.

[0003] When using the above method to test the imaging consistency of a camera, there are requirements for the testing equipment, the testing environment, and the operation. The method requires a collimator, a light source, and a high-precision micro-rotating stage for adjusting the camera's attitude. During the test, the optical path must be isolated to ensure that the imaging is not interfered with by external light sources. During operation, the high-precision micro-rotating stage must be used to adjust the three axes of the camera separately and record the readings. Finally, the multiple sets of readings from the high-precision micro-rotating stage are processed and converted to obtain the measurement result. When using a collimator for testing, the requirements for the testing equipment are quite stringent, the testing environment has certain requirements, and the precision of each device is also required. In addition, there are alignment errors when adjusting with the micro-rotating stage, and visual errors exist when aligning the crosshair pattern of the collimator. These factors make the collimator testing method a costly, inconvenient, and error-prone testing method.

[0004] Considering the above, how to design and implement a monocular camera imaging consistency detection method that has low requirements for detection equipment, simple operation, is not easily affected by the environment, is inexpensive, and easy to apply has become a key problem that urgently needs to be solved in the field of camera engineering applications. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] The technical problem this invention aims to solve is: in situations where optical instruments similar to collimators are unavailable or high detection accuracy is required, a monocular camera imaging consistency detection device is designed and implemented that is low-requirement, easy to operate, not easily affected by the environment, inexpensive, and convenient to use, thereby completing the imaging consistency detection of a monocular camera; furthermore, based on this imaging consistency detection device, this invention designs and implements a monocular camera imaging consistency detection method based on relative pixel coordinates; this method only requires a checkerboard calibration plate and a strapdown fixing fixture that meet specified requirements, and processes the calibration plate images before and after the camera's impact vibration test to calculate the offset of the camera's principal optical axis, thereby achieving monocular camera imaging consistency detection.

[0007] (II) Technical Solution

[0008] To solve the above-mentioned technical problems, the present invention provides a monocular camera imaging consistency detection device based on relative pixel coordinates. The device includes: strapdown fixing fixture, detection target plate and imaging consistency calculation software.

[0009] The strapdown fixing fixture includes: a monocular camera strapdown fixing fixture and a detection target plate strapdown fixing fixture;

[0010] The detection target plate has a pattern for camera detection on one side. The pattern consists of squares of fixed length arranged in a black and white checkerboard pattern, with the number of rows less than the number of columns. The detection target plate is mounted on a strapdown fixture.

[0011] The monocular camera is mounted on a monocular camera strapdown fixture and connected to an image processing computer via a dedicated cable. Data processing and image consistency calculation are performed using imaging consistency calculation software.

[0012] Furthermore, the present invention also provides a method for monocular camera imaging consistency detection based on relative pixel coordinates, the method being implemented using the aforementioned apparatus, the method comprising:

[0013] The method consists of eight steps: assembly and placement, re-inspection and fine-tuning, image acquisition, image coordinate system construction, relative pixel coordinate calculation, reprojection error analysis and camera optical axis change angle calculation. The basic flowchart of the method is shown in Figure 1.

[0014] The method includes the following steps:

[0015] Step A. Assemble and place the imaging consistency detection device; On a reference plane with a length of not less than 2.5m and a width of not less than 1m, assemble and place the detection target plate, monocular camera, and strapdown fixing fixture in sequence on the reference plane;

[0016] Step B. Perform a re-inspection and fine-tuning of the assembled device; ensure that the installed detection target plate is at a 90-degree angle to the reference plane; ensure that the optical path of the installed monocular camera is parallel to the reference plane, and that the monocular camera is facing the checkerboard pattern of the detection target plate.

[0017] Step C. Power on and debug the device after the re-inspection and fine-tuning; connect the monocular camera to the image processing computer through a dedicated cable, observe the real-time image of the monocular camera through the image processing computer, and ensure that the image is clear and the checkerboard pattern of the detection target is completely in the field of view of the real-time image of the monocular camera after power-on.

[0018] Step D. Provided that Steps A, B, and C are completed, use a monocular camera to acquire images before the flight vibration and shock test. During acquisition, ensure that the checkerboard pattern of the front target plate is basically located in the center of the camera's field of view, occupies more than half of the camera's field of view, and that all checkerboard patterns are within the monocular camera's field of view. If the above conditions are not met, return to Step B. Once image acquisition is completed in this step, all strapdown fixtures must remain stationary and cannot be moved.

[0019] Step E. After completing step D, remove the monocular camera and perform flight vibration and shock tests on it. After the tests, ensure that the camera and the strapdown tooling mounting surface are not damaged.

[0020] Step F. Reinstall the monocular camera, which has completed the flight vibration and shock test, onto the imaging consistency testing device. During this process, ensure that the positions of the strapdown fixture and the test target plate do not change. If the positions change due to improper operation, return to step B and start again.

[0021] Step G. Based on step F, power on the monocular camera again and perform a second image acquisition; since the positions of the strapdown fixture and the detection target plate did not change from step D to step F, the image acquired in step G is the image result after the flight vibration and shock test;

[0022] Step H. Perform checkerboard corner detection on the target board checkerboard image before the flight vibration and shock test in Step D and the target board checkerboard image after the flight vibration and shock test in Step G, respectively.

[0023] Step I. Construct an image pixel coordinate system; with the top left corner of the image as the origin, the image width as the X-axis, and the image height as the Y-axis, with the positive X-axis pointing to the right and the positive Y-axis pointing downwards;

[0024] Step J. Based on Step I, calculate the pixel coordinates of each checkerboard corner point on the two frames of images before and after the flight vibration and shock test;

[0025] Step K. Calculate the reprojection error of the corner detection extracted from the two images before and after the flight vibration and shock test: m, n; if m < 0.5 and n < 0.5, then the two images acquired before and after the test are considered to meet the requirements, and step L can be continued; otherwise, the data acquisition error is considered to be large, and step D is returned to start again.

[0026] Step L. Calculate the relative pixel coordinate changes of each checkerboard corner point in the two images before and after the flight vibration and shock test, and perform mean quantization;

[0027] Step M. Calculate the offset angles of the camera's optical axis in the pitch and yaw directions before and after the flight vibration and shock test;

[0028] Step N. Calculate the offset angle in the roll direction between the two images before and after the flight vibration and shock test; using the known corner pixel coordinates in the two images obtained in step J, calculate the affine transformation between the two images and further calculate the offset angle in the roll direction between the two images before and after the flight vibration and shock test.

[0029] In step H, the checkerboard corner detection is performed using the SURF method, which has high feature point detection accuracy and is widely used in related fields.

[0030] In step J, the method for calculating the pixel coordinates of the chessboard corner points is as follows:

[0031] Step J.1. Based on the detection results of step H, directly extract the pixel coordinates of the checkerboard corner points;

[0032] Step J.2. Combining the image pixel coordinate system constructed in Step I, the corner points of the chessboard extracted in Step J.1 are numbered and sorted sequentially from left to right and from top to bottom.

[0033] Step J.3. List the pixel coordinates of each checkerboard corner point on the two frames of images before and after the flight vibration and shock test, and represent them using the following point set:

[0034] (x0,y0),(x1,y1),(x2,y2)…(x n-1 ,y n-1 )

[0035] and (x'0,y'0),(x'1,y'1),(x'2,y'2)...(x' n-1 ,y' n-1 ).

[0036] In step K, the method for calculating the reprojection error of image corner extraction and detection is as follows:

[0037] Step K.1. Combine the physical length d of the chessboard grid side length, calculate the world coordinate system coordinates of the chessboard grid corner points and project them onto the image plane to obtain the theoretical pixel coordinate values ​​of the chessboard grid corner points;

[0038] Step K.2. Combine with step J to obtain the actual pixel coordinates of the corner points of the chessboard grid;

[0039] Step K.3. For each corner point of the chessboard, calculate the Euclidean distance between the actual pixel coordinates and the theoretical pixel coordinates;

[0040] Step K.4. Take the average of the Euclidean distances calculated for each corner point in a frame of an image to obtain the reprojection error of that frame of the image; thereby calculate the reprojection errors of the corner points of the two images before and after the flight vibration and shock test: β1, β2.

[0041] In step L, the method for calculating the relative pixel coordinate changes and their mean values ​​between the two images is as follows:

[0042] Step L.1. Based on the completion of step J.3, calculate the difference between the pixel coordinates of the corner points of the image before the experiment and the corner points of the image after the experiment:

[0043] (Δx0,Δy0),(Δx1,Δy1),(Δx2,Δy2)…(Δx n-1 ,Δy n-1 );

[0044] Step L.2. Take the average of the relative pixel coordinate changes of the two images obtained in L.1:

[0045]

[0046] In step M, the method for calculating the offset angles of the camera's optical axis in the pitch and yaw directions before and after the flight vibration and shock test is as follows:

[0047] Step M.1. Considering that the camera optical axis change is a small angle change, the angle offset of the camera's main optical axis in pitch and yaw can be considered decoupled and can be derived separately.

[0048] Step M.2. Based on Step M.1, the basic parameters required for derivation include the camera CMOS pixel size Ψ and the focal length f of the monocular camera to be tested; the basic variables required for derivation include the changes in the relative pixel coordinates of the two images before and after the experiment along the X and Y axes. The camera's optical axis is offset by an angle θ in the yaw and pitch directions. x θ y ;

[0049] Step M.3. Construct the relationship between pixel coordinates and optical axis deviation angle; considering that the camera's optical axis variation is a small angular change, the change in X-axis pixel coordinates can be derived. The angle θ between the camera's optical axis and the yaw direction x Relationship:

[0050]

[0051] Similarly, the change in the Y-axis pixel coordinate can be derived. The angle θ between the camera's optical axis and the camera's optical axis in the pitch direction y Relationship:

[0052]

[0053] (III) Beneficial Effects

[0054] Compared with the prior art, the present invention has the following advantages:

[0055] (1) The present invention designs a device that can detect the consistency of monocular camera imaging. This device can complete the detection of monocular camera imaging consistency without optical instruments such as collimators. Compared with the traditional method of detection using optical instruments, this method has low requirements for detection equipment, is simple to operate, is not easily affected by the environment, and is low in cost and easy to apply. It greatly reduces the technical threshold and detection difficulty of monocular camera imaging consistency detection.

[0056] (2) Based on the device, the present invention proposes a monocular camera imaging consistency detection method based on relative pixel coordinates. This method does not require high-end detection equipment. It only requires the use of a specified detection target plate and strapdown fixing fixture to collect images of the detection target plate before and after impact and vibration. The offset angle of the camera's principal optical axis in the pitch and yaw directions and the offset angle of the image in the roll direction can be cleverly calculated by the change in relative pixel coordinates. This method is the first to realize camera imaging consistency detection based on relative pixel coordinates, filling the development vacuum in related fields.

[0057] (3) The present invention has been verified by a large number of experiments. The device and method of the invention have high detection accuracy when performing camera imaging consistency detection. A large amount of experimental data shows that when performing monocular camera imaging consistency detection by the present invention, the angle error is less than 0.01°, which is better than the optical detection method and fully meets the needs of the field of camera imaging consistency detection. Attached Figure Description

[0058] Figure 1 is a flowchart of the technical solution of the present invention. Detailed Implementation

[0059] To make the objectives, contents, and advantages of the present invention clearer, the specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples.

[0060] To solve the above-mentioned technical problems, the present invention provides a monocular camera imaging consistency detection device based on relative pixel coordinates. The device includes: strapdown fixing fixture, detection target plate and imaging consistency calculation software.

[0061] The strapdown fixing fixture includes: a monocular camera strapdown fixing fixture and a detection target plate strapdown fixing fixture;

[0062] The detection target plate has a pattern for camera detection on one side. The pattern consists of squares of fixed length arranged in a black and white checkerboard pattern, with the number of rows less than the number of columns. The detection target plate is mounted on a strapdown fixture.

[0063] The monocular camera is mounted on a monocular camera strapdown fixture and connected to an image processing computer via a dedicated cable. Data processing and image consistency calculation are performed using imaging consistency calculation software.

[0064] Furthermore, the present invention also provides a method for monocular camera imaging consistency detection based on relative pixel coordinates, the method being implemented using the aforementioned apparatus, the method comprising:

[0065] The method consists of eight steps: assembly and placement, re-inspection and fine-tuning, image acquisition, image coordinate system construction, relative pixel coordinate calculation, reprojection error analysis and camera optical axis change angle calculation. The basic flowchart of the method is shown in Figure 1.

[0066] The method includes the following steps:

[0067] Step A. Assemble and place the imaging consistency detection device; On a reference plane with a length of not less than 2.5m and a width of not less than 1m, assemble and place the detection target plate, monocular camera, and strapdown fixing fixture in sequence on the reference plane;

[0068] Step B. Perform a re-inspection and fine-tuning of the assembled device; ensure that the installed detection target plate is at a 90-degree angle to the reference plane; ensure that the optical path of the installed monocular camera is parallel to the reference plane, and that the monocular camera is facing the checkerboard pattern of the detection target plate.

[0069] Step C. Power on and debug the device after the re-inspection and fine-tuning; connect the monocular camera to the image processing computer through a dedicated cable, observe the real-time image of the monocular camera through the image processing computer, and ensure that the image is clear and the checkerboard pattern of the detection target is completely in the field of view of the real-time image of the monocular camera after power-on.

[0070] Step D. Provided that Steps A, B, and C are completed, use a monocular camera to acquire images before the flight vibration and shock test. During acquisition, ensure that the checkerboard pattern of the front target plate is basically located in the center of the camera's field of view, occupies more than half of the camera's field of view, and that all checkerboard patterns are within the monocular camera's field of view. If the above conditions are not met, return to Step B. Once image acquisition is completed in this step, all strapdown fixtures must remain stationary and cannot be moved.

[0071] Step E. After completing step D, remove the monocular camera and perform flight vibration and shock tests on it. After the tests, ensure that the camera and the strapdown tooling mounting surface are not damaged.

[0072] Step F. Reinstall the monocular camera, which has completed the flight vibration and shock test, onto the imaging consistency testing device. During this process, ensure that the positions of the strapdown fixture and the test target plate do not change. If the positions change due to improper operation, return to step B and start again.

[0073] Step G. Based on step F, power on the monocular camera again and perform a second image acquisition; since the positions of the strapdown fixture and the detection target plate did not change from step D to step F, the image acquired in step G is the image result after the flight vibration and shock test;

[0074] Step H. Perform checkerboard corner detection on the target board checkerboard image before the flight vibration and shock test in Step D and the target board checkerboard image after the flight vibration and shock test in Step G, respectively.

[0075] Step I. Construct an image pixel coordinate system; with the top left corner of the image as the origin, the image width as the X-axis, and the image height as the Y-axis, with the positive X-axis pointing to the right and the positive Y-axis pointing downwards;

[0076] Step J. Based on Step I, calculate the pixel coordinates of each checkerboard corner point on the two frames of images before and after the flight vibration and shock test;

[0077] Step K. Calculate the reprojection error of the corner detection extracted from the two images before and after the flight vibration and shock test: m, n; if m < 0.5 and n < 0.5, then the two images acquired before and after the test are considered to meet the requirements, and step L can be continued; otherwise, the data acquisition error is considered to be large, and step D is returned to start again.

[0078] Step L. Calculate the relative pixel coordinate changes of each checkerboard corner point in the two images before and after the flight vibration and shock test, and perform mean quantization;

[0079] Step M. Calculate the offset angles of the camera's optical axis in the pitch and yaw directions before and after the flight vibration and shock test;

[0080] Step N. Calculate the offset angle in the roll direction between the two images before and after the flight vibration and shock test; using the known corner pixel coordinates in the two images obtained in step J, calculate the affine transformation between the two images and further calculate the offset angle in the roll direction between the two images before and after the flight vibration and shock test.

[0081] In step H, the checkerboard corner detection is performed using the SURF method, which has high feature point detection accuracy and is widely used in related fields.

[0082] In step J, the method for calculating the pixel coordinates of the chessboard corner points is as follows:

[0083] Step J.1. Based on the detection results of step H, directly extract the pixel coordinates of the checkerboard corner points;

[0084] Step J.2. Combining the image pixel coordinate system constructed in Step I, the corner points of the chessboard extracted in Step J.1 are numbered and sorted sequentially from left to right and from top to bottom.

[0085] Step J.3. List the pixel coordinates of each checkerboard corner point on the two frames of images before and after the flight vibration and shock test, and represent them using the following point set:

[0086] (x0,y0),(x1,y1),(x2,y2)…(x n-1 ,y n-1 )

[0087] and (x'0,y'0),(x'1,y'1),(x'2,y'2)...(x' n-1 ,y' n-1 ).

[0088] In step K, the method for calculating the reprojection error of image corner extraction and detection is as follows:

[0089] Step K.1. Combine the physical length d of the chessboard grid side length, calculate the world coordinate system coordinates of the chessboard grid corner points and project them onto the image plane to obtain the theoretical pixel coordinate values ​​of the chessboard grid corner points;

[0090] Step K.2. Combine with step J to obtain the actual pixel coordinates of the corner points of the chessboard grid;

[0091] Step K.3. For each corner point of the chessboard, calculate the Euclidean distance between the actual pixel coordinates and the theoretical pixel coordinates;

[0092] Step K.4. Take the average of the Euclidean distances calculated for each corner point in a frame of an image to obtain the reprojection error of that frame of the image; thereby calculate the reprojection errors of the corner points of the two images before and after the flight vibration and shock test: β1, β2.

[0093] In step L, the method for calculating the relative pixel coordinate changes and their mean values ​​between the two images is as follows:

[0094] Step L.1. Based on the completion of step J.3, calculate the difference between the pixel coordinates of the corner points of the image before the experiment and the corner points of the image after the experiment:

[0095] (Δx0,Δy0),(Δx1,Δy1),(Δx2,Δy2)…(Δx n-1 ,Δy n-1 );

[0096] Step L.2. Take the average of the relative pixel coordinate changes of the two images obtained in L.1:

[0097]

[0098] In step M, the method for calculating the offset angles of the camera's optical axis in the pitch and yaw directions before and after the flight vibration and shock test is as follows:

[0099] Step M.1. Considering that the camera optical axis change is a small angle change, the angle offset of the camera's main optical axis in pitch and yaw can be considered decoupled and can be derived separately.

[0100] Step M.2. Based on Step M.1, the basic parameters required for derivation include the camera CMOS pixel size Ψ and the focal length f of the monocular camera to be tested; the basic variables required for derivation include the changes in the relative pixel coordinates of the two images before and after the experiment along the X and Y axes. The camera's optical axis is offset by an angle θ in the yaw and pitch directions. x θ y ;

[0101] Step M.3. Construct the relationship between pixel coordinates and optical axis deviation angle; considering that the camera's optical axis variation is a small angular change, the change in X-axis pixel coordinates can be derived. The angle θ between the camera's optical axis and the yaw direction x Relationship:

[0102]

[0103] Similarly, the change in the Y-axis pixel coordinate can be derived. The angle θ between the camera's optical axis and the camera's optical axis in the pitch direction y Relationship:

[0104]

[0105] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting the consistency of images from a monocular camera based on relative pixel coordinates, characterized in that, The method is implemented based on a monocular camera imaging consistency detection device, which includes: a strapdown fixture, a detection target plate, and imaging consistency calculation software. The strapdown fixture includes a monocular camera strapdown fixture and a detection target plate strapdown fixture. The detection target plate has a pattern for camera detection on one side. The pattern consists of squares of fixed length arranged in a black and white checkerboard pattern, with the number of rows less than the number of columns. The detection target plate is mounted on the detection target plate strapdown fixture. The monocular camera is mounted on the monocular camera strapdown fixture and connected to an image processing computer via a dedicated cable. Data processing and image consistency calculation are performed using the imaging consistency calculation software. The method includes eight steps: assembly and placement of the monocular camera imaging consistency static detection device, re-inspection and fine-tuning of the monocular camera imaging consistency static detection device, monocular camera image acquisition, monocular camera image coordinate system construction, monocular camera relative pixel coordinate calculation, monocular camera reprojection error analysis, and monocular camera optical axis change angle calculation. The method includes the following steps: Step A. Assemble and place the imaging consistency detection device; on a reference plane with a length of not less than 2.5m and a width of not less than 1m, assemble and place the detection target plate, monocular camera, and strapdown fixing fixture in sequence on the reference plane; Step B. Step C. Perform a final inspection and fine-tuning of the assembled device; ensure that the installed target plate is at a 90-degree angle to the reference plane; ensure that the optical path of the installed monocular camera is parallel to the reference plane, and that the monocular camera is directly facing the checkerboard pattern of the target plate; Step D. Power on and debug the device after the final inspection and fine-tuning; connect the monocular camera to the image processing computer via a dedicated cable, and observe the real-time image of the monocular camera through the image processing computer to ensure that the image is clear and that the checkerboard pattern of the target plate is completely within the real-time image field of view of the monocular camera after power-on; Step D. On the premise that steps A, B, and C are completed, use the monocular camera to acquire images before the flight vibration and shock test; during acquisition, it is necessary to ensure that the checkerboard pattern of the front target plate is basically located in the center of the camera's field of view, occupies more than half of the camera's field of view, and that the entire checkerboard pattern is within the field of view of the monocular camera; If the above conditions are not met, return to step B. Once image acquisition is completed in this step, all strapdown fixtures must remain stationary and cannot be moved. Step E. After step D, remove the monocular camera and perform flight vibration and shock tests. After the tests, ensure that the camera and the mounting surface of the strapdown fixture are not damaged. Step F. Reinstall the monocular camera, after the flight vibration and shock tests, onto the imaging consistency detection device. During this process, ensure that the positions of the strapdown fixture and the detection target plate do not change. If the positions change due to improper operation, return to step B and start again. Step G. Based on step F, power on the monocular camera again and perform a second image acquisition. Since the positions of the strapdown fixture and the detection target plate did not change from step D to step F, the image acquired in step G is the image result after the flight vibration and shock test. Step H.Step D: Detect corner points on the target board checkerboard image before the flight vibration and impact test and the target board checkerboard image after the flight vibration and impact test, respectively. Step I: Construct an image pixel coordinate system; with the upper left corner of the image as the origin, the image width as the X-axis, and the image height as the Y-axis, with the positive X-axis pointing to the right and the positive Y-axis pointing downwards. Step J: Based on Step I, calculate the pixel coordinates of each checkerboard corner point on the two frames of images before and after the flight vibration and impact test. Step K: Calculate the reprojection error of the corner point detection extraction from the two images before and after the flight vibration and impact test. , ;if and If the two images acquired before and after the test meet the requirements, proceed to step L; otherwise, if the data acquisition error is considered large, return to step D and start again. Step L: Calculate the relative pixel coordinate changes of each checkerboard corner point in the two images before and after the flight vibration and shock test and perform mean quantization. Step M: Calculate the offset angle of the camera optical axis in the pitch and yaw directions before and after the flight vibration and shock test. Step N: Calculate the offset angle of the two images in the roll direction before and after the flight vibration and shock test. Using the known corner pixel coordinates in the two images obtained in step J, calculate the affine transformation between the two images and further calculate the offset angle of the two images in the roll direction before and after the flight vibration and shock test.

2. The monocular camera imaging consistency detection method based on relative pixel coordinates as described in claim 1, characterized in that, In step H, the corner point detection of the chessboard is performed using the SURF method.

3. The monocular camera imaging consistency detection method based on relative pixel coordinates as described in claim 2, characterized in that, In step J, the method for calculating the pixel coordinates of the corner points of the chessboard is as follows: Step J.

1. Combine the detection results of step H and directly extract the pixel coordinates of the corner points of the chessboard. Step J.

2. Using the image pixel coordinate system established in Step I, number and sort the checkerboard corner points extracted in Step J.1 sequentially from left to right and from top to bottom; Step J.

3. List the pixel coordinates of each checkerboard corner point on the two frames of images before and after the flight vibration and impact test, and represent them using the following point set: , , … and , , … 。 4. The monocular camera imaging consistency detection method based on relative pixel coordinates as described in claim 3, characterized in that, In step K, the method for calculating the reprojection error of image corner point extraction and detection is as follows: Step K.

1. Combine the physical length d of the chessboard grid side length to calculate the world coordinate system coordinates of the chessboard grid corner points and project them onto the image plane to obtain the theoretical pixel coordinate values ​​of the chessboard grid corner points; Step K.

2. Combine with step J to obtain the actual pixel coordinate values ​​of the chessboard grid corner points; Step K.

3. For each chessboard grid corner point, calculate the Euclidean distance between the actual pixel coordinates and the theoretical pixel coordinates; Step K.

4. Take the average of the Euclidean distances calculated for each corner point in a frame of image to obtain the reprojection error of that frame of image; thus, the reprojection error of the corner points calculated in the two images before and after the flight vibration and shock test is calculated. , 。 5. The monocular camera imaging consistency detection method based on relative pixel coordinates as described in claim 4, characterized in that, In step L, the method for calculating the relative pixel coordinate changes and their mean values ​​between the two images is as follows: Step L.

1. Based on the completion of step J.3, the pixel coordinates of the corner points of the image before the experiment and the corner points of the image after the experiment are subtracted respectively: , , … ； Step L.

2. Take the average of the relative pixel coordinate changes of the two images obtained in L.1: 。 6. The monocular camera imaging consistency detection method based on relative pixel coordinates as described in claim 5, characterized in that, In step M, the method for calculating the offset angles of the camera optical axis in the pitch and yaw directions before and after the flight vibration and shock test is as follows: Step M.

1. Considering that the camera optical axis change is a small angle change, the offset angles of the camera principal optical axis in the pitch and yaw directions are considered decoupled and derived separately; Step M.

2. Under the premise of step M.1., the basic parameters required for derivation include the camera CMOS pixel size. and the focal length of the monocular camera under test The basic variables required for the derivation include the changes in the relative pixel coordinates of the two images before and after the experiment along the X and Y axes. The angle of offset of the camera's optical axis in the yaw and pitch directions ， ； Step M.

3. Construct the relationship between pixel coordinates and optical axis deviation angle; considering that the camera's optical axis variation is a small angle change, derive the change in X-axis pixel coordinates. Angle of offset from the camera's optical axis in the yaw direction Relationship: Similarly, the change in the Y-axis pixel coordinate can be derived. Angle of offset from the camera's optical axis in the pitch direction Relationship: 。

Images