A high-altitude visual measurement system calibration method, device and readable storage medium

Abstract

The application discloses a high-altitude visual measurement system calibration method and device and a readable storage medium. The method comprises the following steps: acquiring a first calibration result, wherein the first calibration result is based on a calibration result when a target is placed on the ground; after the first calibration is completed, a first scene picture is collected by using a binocular camera; the pitch angle of the camera is increased, and an image taken by aiming at a predetermined high-altitude area is recorded as a second scene picture; a pitch angle change value is determined according to the difference between pixel point data of feature points of a feature object in the first scene picture and pixel point data of the feature points in the second scene picture; a second calibration result is determined based on the first calibration result and the pitch angle change value, wherein the second calibration result represents a calibration parameter when the binocular camera takes an image by aiming at the high-altitude area. The application sets a calibration target at a low altitude to realize visual calibration of a high-altitude scene, and does not need to mount an RTK calibration on a UAV, thereby effectively reducing system cost and operation complexity.

Description

Technical Field

[0001] This invention relates to the field of camera calibration technology, and specifically to a calibration method, apparatus, and readable storage medium for an aerial vision measurement system. Background Technology

[0002] In fields such as computer vision, remote sensing and mapping, and UAV autonomous navigation, the accuracy of camera parameters is crucial as the core perception sensor. Camera calibration is the process of calculating and determining the camera's internal parameters (such as focal length, principal point coordinates, distortion coefficients, etc.) and external parameters (such as position and attitude). Its accuracy directly determines the success or failure of subsequent vision tasks (such as 3D reconstruction, target localization, and stereo vision measurement). For near-ground or indoor scenes, calibration technology is quite mature, and it is usually achieved by photographing a high-precision calibration board of known size (such as a checkerboard or dot array) in a controlled environment.

[0003] However, when applications extend to altitudes of tens or even hundreds of meters, traditional calibration methods face severe challenges. The wide field of view at high altitudes makes it difficult to deploy and identify large-scale physical calibration boards that meet accuracy requirements in the natural environment. Even if they could be deployed, the manufacturing cost, transportation, and deployment difficulties are extremely high. Therefore, traditional short-range calibration board methods are almost impossible to apply directly in high-altitude scenarios.

[0004] To address this challenge, existing technologies employ RTK calibration mounted on UAVs. However, this calibration method is costly, and high-precision RTK significantly increases the overall hardware overhead. Furthermore, the system integration and operation are highly complex, requiring not only precise mechanical installation to ensure stable relative relationships between sensors but also addressing stringent time synchronization issues. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides a calibration method, apparatus, and readable storage medium for a high-altitude visual measurement system. This eliminates the need for RTK-mounted UAVs for calibration, effectively improving the efficiency of high-altitude calibration and reducing system costs. The technical solution includes: Firstly, a calibration method for a high-altitude visual measurement system is provided, comprising the following steps: Obtain the first calibration result, which is the calibration result of the binocular camera obtained by taking images of the ground target with the binocular camera when the target is placed on the ground; the field of view space when the binocular camera takes images of the ground target is recorded as the ground field of view space; After the first calibration is completed, a scene image of the ground field of view is captured using a binocular camera and recorded as the first scene image. The binocular camera is kept stationary and does not rotate horizontally. The camera's pitch angle is increased so that the binocular camera's shooting range is aligned with a predetermined high-altitude area. The image captured with the binocular camera aligned with the predetermined high-altitude area is recorded as the second scene image. The field of view of the high-altitude area captured by the binocular camera is recorded as the high-altitude field of view space. The bottom area of ​​the high-altitude field of view space coincides with the top area of ​​the ground field of view space, and the overlapping area contains a feature object. Based on the difference between the pixel data of the feature points of the feature object in the first scene image and the pixel data in the second scene image, the pitch angle change value of the binocular camera from capturing the first scene image to capturing the second scene image is determined. Based on the first calibration result and the pitch angle change value, a second calibration result is determined. The second calibration result represents the calibration parameters when the binocular camera is aimed at a high-altitude area for shooting.

[0006] In some implementations, the method for obtaining the first calibration result includes: Based on the four coordinate values ​​of the target point in the world coordinate system, camera coordinate system, image coordinate system, and pixel coordinate system, a first model representing the transformation relationship from the pixel coordinate system to the world coordinate system is obtained. The first model includes the camera's intrinsic parameter matrix and extrinsic parameter matrix, and the extrinsic parameter matrix includes a rotation matrix and a translation matrix. The camera is calibrated using the acquired target images to obtain the intrinsic and extrinsic parameter matrices. A second model is obtained to represent the relationship between the extrinsic parameter matrix parameters and the first parameter. The value of the first parameter is determined using the determined extrinsic parameter matrix parameters. The first parameter includes the rotation angle between the world coordinate system and the camera coordinate system. The rotation angle includes the rotation angle with the three coordinate axes as the rotation center. The second model includes the first relationship between the rotation matrix and the rotation angle, and also includes the second relationship between the coordinates of any point in space in the camera coordinate system and the coordinates in the world coordinate system, and the rotation matrix and the translation matrix.

[0007] In some implementations, obtaining the first model representing the transformation relationship from the pixel coordinate system to the world coordinate system based on the four coordinate values ​​of the target point in the world coordinate system, camera coordinate system, image coordinate system, and pixel coordinate system includes: Let the matrix be... ;in, These are the target's coordinates in the pixel coordinate system; These are the coordinates of the target in the world coordinate system; K is the intrinsic parameter matrix, R is the rotation matrix, and T is the translation matrix.

[0008] In some implementations, the second model representing the relationship between the extrinsic parameter matrix parameters and the first parameter is: Camera rotation matrix for: , This represents the value of the element in the i-th row and j-th column of the matrix. It is related to the trigonometric function value of the rotation angle between the world coordinate system and the corresponding camera coordinate system; Camera translation matrix for: ;in, The coordinates of the target point in the world coordinate system. The coordinates of the target point in the camera coordinate system.

[0009] In some implementations, determining the second calibration result based on the first calibration result and the pitch angle change value includes: Replace the pitch angle in the extrinsic parameter matrix represented by the first parameter in the first calibration result with the sum of the pitch angle and the pitch angle change value in the first calibration result; Obtain the updated extrinsic parameter matrix; The second calibration result is based on the intrinsic parameter matrix and the updated extrinsic parameter matrix from the first calibration result.

[0010] In some implementations, determining the pitch angle change value of the binocular camera from capturing the first scene image to capturing the second scene image based on the difference between the pixel data of the feature points of the feature object in the first scene image and the pixel data in the second scene image includes: The pixel position of the feature point of the feature object in the first scene image is recorded as the first pixel position. Obtain the pixel position of the feature point in the second scene image, and denote it as the second pixel position; The pitch angle change value is determined based on the positions of the first and second pixels.

[0011] In some implementations, the step of calibrating the camera using the acquired target images to obtain intrinsic and extrinsic parameter matrices includes: After obtaining the coordinates of the target point in the world coordinate system and the pixel coordinate system, a normalization transformation is performed on the coordinates in the world coordinate system and the pixel coordinate system to eliminate the difference in dimensions. The normalization method for the pixel coordinates of the target point includes the following steps: Get the pixel coordinates of all target points ( ), i = 1, 2, ..., n, where n is the number of target points; Calculate the centroid coordinates of all points, i.e., the average position of all points. ); Subtract the average position from the pixel coordinates of all target points so that the centroid of all points moves to the origin after each point is translated, thus obtaining the updated pixel coordinates of the target points. ); Calculate the average Euclidean distance from all points to the origin after translation. ; Calculate scaling factor , When the coordinates of a point are two-dimensional coordinates When the coordinates of a point are three-dimensional coordinates ; After translation, scale all points to obtain normalized coordinates. ), ; The normalization method for the world coordinate system of the target point is the same as the normalization method for the pixel coordinates of the target point described above.

[0012] In some implementations, the step of calibrating the camera using the acquired target images to obtain intrinsic and extrinsic parameter matrices includes: Based on the collected target images, the pixel coordinates of multiple target points are obtained. Combined with the world coordinates of the target points and the first model, the H matrix is ​​solved. For the H matrix, the RQ decomposition is used to decompose the H matrix to obtain the intrinsic parameter matrix and the extrinsic parameter matrix; The optimal solutions for the intrinsic and extrinsic parameters are obtained by optimizing the matrix parameters using the LM algorithm.

[0013] Secondly, a calibration device for a high-altitude visual measurement system is provided, the device comprising: The first calibration unit is used to obtain the first calibration result, which is the calibration result of the binocular camera obtained by taking images of the ground target with the binocular camera when the target is placed on the ground; the field of view space when the binocular camera takes images of the ground target is recorded as the ground field of view space; The low-altitude and high-altitude imaging unit is used to acquire scene images of the ground-level field of view using a binocular camera after the first calibration is completed, and these images are recorded as the first scene images. The binocular camera is kept stationary and does not rotate horizontally, while the camera's pitch angle is increased so that the binocular camera's shooting range is aligned with a predetermined high-altitude area. The images captured in alignment with the predetermined high-altitude area are recorded as the second scene images. The field of view of the binocular camera when shooting the high-altitude area is recorded as the high-altitude field of view space. The bottom area of ​​the high-altitude field of view space overlaps with the top area of ​​the ground-level field of view space, and the overlapping area contains feature objects. Angle change determination unit is used to determine the pitch angle change value of the binocular camera from shooting the first scene image to shooting the second scene image based on the difference between the pixel data of the feature points of the feature object in the first scene image and the pixel data in the second scene image. The second calibration unit is used to determine the second calibration result based on the first calibration result and the pitch angle change value. The second calibration result represents the calibration parameters when the binocular camera is aimed at a high-altitude area for shooting.

[0014] Thirdly, a computer-readable storage medium is provided having computer instructions stored thereon, which, when executed by a processor, implement the steps of the high-altitude visual measurement system calibration method as described in the first aspect above.

[0015] The present invention provides a calibration method, apparatus, and readable storage medium for a high-altitude visual measurement system, which has the following advantages: The present invention utilizes a calibration target set at low altitude for calibration, obtaining a first calibration result at low altitude. Then, by adding a pitch angle to the first calibration result, a second calibration result is obtained when facing high altitude. The present invention achieves visual calibration of high-altitude scenes using only a calibration target set at low altitude, eliminating the need for RTK calibration on a UAV, effectively reducing system cost and operational complexity. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a calibration method for an aerial vision measurement system according to an embodiment of this application; Figure 2 This is a schematic diagram of a scenario for a calibration method of an aerial vision measurement system according to an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a calibration device for a high-altitude visual measurement system according to an embodiment of this application. Detailed Implementation

[0017] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0018] See Figure 1 This application provides a calibration method for an aerial vision measurement system, comprising the following steps: Step 1: Obtain the first calibration result. The first calibration result is the calibration result of the binocular camera obtained by taking images of the ground target with the binocular camera when the target is placed on the ground. The field of view space when the binocular camera takes images of the ground target is recorded as the ground field of view space. Step 2: After the first calibration is completed, use a binocular camera to capture scene images of the ground view space, and record them as the first scene images; keep the binocular camera from translating or rotating horizontally, and increase the camera's pitch angle so that the binocular camera's shooting range is aligned with the predetermined high-altitude area, and record the images captured aligned with the predetermined high-altitude area as the second scene images; record the view space of the high-altitude area captured by the binocular camera as the high-altitude view space, the bottom area of ​​the high-altitude view space and the top area of ​​the ground view space overlap, and the overlapping area contains feature objects, such as mountain peaks or plants, which can be captured by both the low-altitude shooting scene and the high-altitude shooting scene simultaneously; Step 3: Based on the difference between the pixel data of the feature points of the feature objects in the first scene image and the pixel data in the second scene image, determine the pitch angle change value of the binocular camera from capturing the first scene image to capturing the second scene image. Step 4: Based on the first calibration result and the pitch angle change value, determine the second calibration result. The second calibration result represents the calibration parameters when the binocular camera is aimed at the high-altitude area for shooting.

[0019] In this embodiment, calibration is performed using a low-altitude calibration target to obtain a first calibration result at low altitude. Then, by adding a pitch angle to the first calibration result, a second calibration result is obtained when facing high altitude. In this embodiment, visual calibration of high-altitude scenes can be achieved using only a low-altitude calibration target, eliminating the need for RTK calibration on a UAV, effectively reducing system cost and operational complexity.

[0020] Specifically, see Figure 2 Wherein, ① and ② are turntables below the camera, used to control the camera's tilt angle, and ③ and ④ are the camera itself. The steps in this embodiment are as follows: First, point the binocular camera at the ground area to take pictures. The visual intersection space formed by the binocular camera is area ABCD. Set up a calibration target in this area, perform large field of view calibration of the camera (including internal and external parameters), and acquire images of the ground area. Then raise the camera's pitch angle, while keeping the horizontal and roll angles unchanged. The visual intersection space formed by the binocular cameras is region A1B1C1D1. It can be understood that the top of region ABCD and the bottom of region A1B1C1D1 intersect, and this intersection area is region DEB1F in the figure. By utilizing the static feature points of characteristic objects within the DEB1F region (such as mountain peaks or plants), and based on the pixel position changes of the same feature points before and after pitch, pitch changes are calculated inversely, calibration parameters are converted, and then calibration parameters for high-altitude shooting (after pitch is raised) are generated.

[0021] In one embodiment, the method for obtaining the first calibration result in step 1 above includes: Step 101: Based on the four coordinate values ​​of the target point in the world coordinate system, camera coordinate system, image coordinate system, and pixel coordinate system, obtain a first model representing the transformation relationship from the pixel coordinate system to the world coordinate system. The first model includes the camera's intrinsic parameter matrix and extrinsic parameter matrix. The extrinsic parameter matrix includes a rotation matrix and a translation matrix. Step 102: Use the acquired target images to calibrate the camera and obtain the intrinsic and extrinsic parameter matrices. Step 103: Obtain a second model representing the relationship between the extrinsic parameter matrix parameters and the first parameter. Using the determined extrinsic parameter matrix parameters, determine the value of the first parameter. The first parameter includes the rotation angle between the world coordinate system and the camera coordinate system. The rotation angle includes the rotation angle around the three coordinate axes as rotation centers. The second model includes the first relationship between the rotation matrix and the rotation angle, and also includes the second relationship between the coordinates of any point in space in the camera coordinate system and the coordinates in the world coordinate system, and the rotation matrix and the translation matrix.

[0022] In this embodiment of the application, two expression models are used to express the rotation matrix in steps 101-103 above. One is of the form of ,in , ... One is the calibration parameter to be solved; the other is a rotation matrix containing rotation angle parameters. This is determined using acquired target images. , ... Then, the rotation angle values ​​in the rotation matrix expressed as rotation angle parameters can be determined.

[0023] For ease of calculation, in step 103, the origin of the camera coordinate system is selected as "any point in space". The coordinates of the "any point in space" in the camera coordinate system are (0, 0, 0). In the first relationship, the values ​​of each element in the rotation matrix are expressed as a function of the rotation angle parameter.

[0024] In some implementations, step 101 above, based on the four coordinate values ​​of the target point in the world coordinate system, camera coordinate system, image coordinate system, and pixel coordinate system, obtains a first model representing the transformation relationship from the pixel coordinate system to the world coordinate system, including: Let the matrix be... ;in, These are the target's coordinates in the pixel coordinate system; K is the target's coordinates in the world coordinate system; R is the intrinsic parameter matrix, T is the rotation matrix, and T is the translation matrix.

[0025] Generally, based on the pinhole imaging model, the transformation matrix from the pixel coordinate system to the world coordinate system can be obtained. Assuming the coordinates of a point in space in the world coordinate system are... The coordinates in the camera coordinate system are The coordinates in the image coordinate system are The coordinates in the pixel coordinate system are The transformation relationship between the image coordinate system and the pixel coordinate system is as follows: ;in, The coordinates of the origin of the image coordinate system in the pixel coordinate system; For the width of a single photosensitive element, For the high efficiency of a single photosensitive element; The transformation relationship between the camera coordinate system and the image coordinate system is as follows: ; The transformation relationship between the world coordinate system and the camera coordinate system is as follows: ;Right now ; but: The transformation relationship from pixel coordinate system to world coordinate system is as follows: ; in, For camera focal length, For the camera intrinsic parameter matrix, , These are the rotation and translation matrices of the extrinsic parameter matrix, respectively.

[0026] In one implementation, step 103 above, wherein the second model characterizing the relationship between the extrinsic parameter matrix parameters and the first parameter is: Camera rotation matrix for: , This represents the value of the element in the i-th row and j-th column of the matrix. It is related to the trigonometric function value of the rotation angle between the world coordinate system and the corresponding camera coordinate system; Camera translation matrix for: ;in, The coordinates of the target point in the world coordinate system. Let be the coordinates of the target point in the camera coordinate system. In one implementation, the target point is taken as the origin of the camera coordinate system, then the translation matrix of the camera in the second model is... for: .

[0027] Generally, the origin of the pixel coordinate system is located at the upper left corner of the image, and the two coordinate axes of the pixel coordinate system are parallel to the image plane, or even parallel to the two sides of the image plane. The origin of the image coordinate system is located at the intersection of the optical axis and the image plane, and the two coordinate axes of the image coordinate system are parallel to the image plane, or even parallel to the two sides of the image plane. The origin of the camera coordinate system is located at the optical center of the lens, the x-axis and y-axis are parallel to the two sides of the image plane, and the Z-axis is the optical axis of the lens, perpendicular to the image plane. The world coordinate system can be freely determined according to the actual situation.

[0028] Assuming the world coordinate system is rotated by an angle A around the Z-axis (counterclockwise), then by an angle B around the Y-axis, and then by an angle E around the X-axis, the camera coordinate system is obtained after three rotations. Then: The rotation matrix P1 about the Z-axis is: ; The rotation matrix P2 about the Y-axis is: ; The rotation matrix P3 about the X-axis is: ; The rotation matrix from the world coordinate system to the camera coordinate system for: ; In one implementation, if the position of the world coordinate system can be obtained by rotating the world coordinate system twice or once to obtain the camera coordinate system, then the rotation matrix from the world coordinate system to the camera coordinate system is... It can be further simplified. For example, if the world coordinate system can be obtained by rotating it by an angle A around the Z-axis and an angle E around the X-axis (i.e., angle B is 0 degrees), then the rotation matrix from the world coordinate system to the camera coordinate system is... for: ; In one implementation, if one camera in a stereo camera is denoted as camera C1 and the other as camera C2, then the rotation matrix from the world coordinate system to the camera coordinate system of camera C1 is... for: in, , , These are the angles of rotation around the Z-axis, the Y-axis, and the X-axis in the camera coordinate system of camera C1 obtained by rotating the world coordinate system. Rotation matrix from world coordinate system to camera coordinate system C2 Similarly, we can obtain the following.

[0029] Obtain the rotation matrix from the world coordinate system to the camera coordinate system C1. Rotation matrix from the world coordinate system to the camera coordinate system C2 Then, further, the translation matrix from the world coordinate system to the camera coordinate system of camera C1 can be obtained. ,in The coordinates of the target point in the world coordinate system. Given the coordinates of the target point in the camera coordinate system of camera C1, we can similarly obtain the translation matrix from the world coordinate system to the camera coordinate system of camera C2. .

[0030] In one implementation, step 4, determining the second calibration result based on the first calibration result and the pitch angle change value, includes: Step 401: Replace the pitch angle in the extrinsic parameter matrix represented by the first parameter in the first calibration result with the sum of the pitch angle and the pitch angle change value in the first calibration result; Step 402: Obtain the updated extrinsic parameter matrix; Step 403: Use the intrinsic parameter matrix and the updated extrinsic parameter matrix from the first calibration result as the second calibration result.

[0031] It should be noted that the pitch angle change represents the change in the camera's field of view in the height direction. The pitch angle change is determined based on the rotation angle of the camera coordinate system in the extrinsic parameter matrix around one of the three coordinate axes. A change in the camera's pitch angle indicates that the camera coordinate system has rotated around one of these axes, and the pitch angle change value can be obtained based on this rotation angle. The extrinsic parameter matrix contains rotation angle parameters around the Z-axis, Y-axis, and X-axis. A change in the camera's pitch angle will cause a change in one of the rotation angles in the extrinsic parameter matrix.

[0032] Commonly, the camera coordinate system is set as follows: the origin of the camera coordinate system is located at the optical center of the lens, the x-axis and y-axis are parallel to the two sides of the image plane, and the Z-axis is the lens optical axis, perpendicular to the image plane. Therefore, if the direction of the side of the image plane parallel to the ground is taken as the x-axis, then the pitch angle in the extrinsic parameter matrix is ​​the angle of rotation of the world coordinate system around the x-axis. Furthermore, the change in pitch angle represents the angle of rotation of the camera coordinate system around the x-axis from before the pitch angle was increased (raised) to after the pitch angle was increased (raised).

[0033] Replace the pitch angle in the extrinsic parameter matrix represented by the first parameter in the first calibration result with the sum of the pitch angle and the pitch angle change value in the first calibration result, that is: Assuming the pitch angle in the extrinsic parameter matrix represented by the first parameter is angle E, the pitch angle change value Then replace the original angle E in the extrinsic parameter matrix with (E+ ).

[0034] In one implementation, step 3 above, which involves determining the pitch angle change value of the binocular camera from capturing the first scene image to capturing the second scene image based on the difference between the pixel data of the feature points of the feature object in the first scene image and the pixel data in the second scene image, includes: The pixel position of the feature point of the feature object in the first scene image is recorded as the first pixel position. Obtain the pixel position of the feature point in the second scene image, and denote it as the second pixel position; The pitch angle change value is determined based on the positions of the first and second pixels.

[0035] Specifically, determining the pitch angle change value based on the positions of the first and second pixels can be achieved by: based on the height of the first pixel position. ,calculate Based on the height of the first pixel position ,calculate The pitch angle change value is - . This refers to the camera's focal length. It should be noted that the pixel position is determined by the camera coordinate system. The origin of the camera coordinate system is located at the optical center of the lens. The x-axis and y-axis are parallel to the two sides of the image plane, respectively, and the Z-axis is the lens optical axis, perpendicular to the image plane.

[0036] In one embodiment, step 102 above, which involves calibrating the camera using the acquired target image to obtain intrinsic and extrinsic parameter matrices, includes: After obtaining the coordinates of the target point in the world coordinate system and the pixel coordinate system, a normalization transformation is performed on the coordinates in the world coordinate system and the pixel coordinate system to eliminate the difference in dimensions. The normalization method includes: normalizing the pixel coordinates of the target point to the average distance from the farthest point. For the world coordinates of the target point, the normalized average distance to the origin is: ; The normalization method for the pixel coordinates of the target point includes the following steps: Get the pixel coordinates of all target points ( ), i = 1, 2, ..., n, where n is the number of target points; Calculate the centroid coordinates of all points, i.e., the average position of all points. ); Subtract the average position from the pixel coordinates of all target points so that the centroid of all points moves to the origin after each point is translated, thus obtaining the updated pixel coordinates of the target points. ); Calculate the average Euclidean distance from all points to the origin after translation. ; Calculate scaling factor , When the coordinates of a point are two-dimensional coordinates When the coordinates of a point are three-dimensional coordinates ; After translation, scale all points to obtain normalized coordinates. ), ; The normalization method for the world coordinate system of the target point is the same as the normalization method for the pixel coordinates of the target point described above.

[0037] Furthermore, a normalized matrix of the pixel coordinates of a point can be constructed. For two-dimensional coordinates, it is used to convert the original coordinates Switch to ; = ; ; Similarly, for three-dimensional world coordinates, a normalized matrix of the world coordinates of a point can be constructed. .

[0038] After normalization, by acquiring the pixel coordinates and world coordinates of multiple target points in the target image, the transformation matrix between pixel coordinates and world coordinates under normalized coordinates can be obtained. Then, the transformation matrix between pixel coordinates and world coordinates under the original coordinate data before normalization. = This solution process does not consider phase difference. Preferably, this application will use the LM optimization algorithm to iteratively solve for the optimal calibration parameters.

[0039] In one embodiment, step 102 above, which involves calibrating the camera using the acquired target image to obtain intrinsic and extrinsic parameter matrices, includes: Based on the collected target images, the pixel coordinates of multiple target points are obtained. Combined with the world coordinates of the target points and the first model, the H matrix is ​​solved. For the H matrix, the RQ decomposition is used to decompose the H matrix to obtain the intrinsic parameter matrix and the extrinsic parameter matrix; The optimal solutions for the intrinsic and extrinsic parameters are obtained by optimizing the matrix parameters using the LM algorithm.

[0040] The objective function used in the LM algorithm is: ,in, This represents the coordinates of a spatial point projected onto the image through intrinsic parameters, extrinsic parameters, and distortion coefficients. This represents the actual coordinates of the spatial point in the image, P is the coordinates of the spatial point in the world coordinate system, and K, R, T, and distcoeffs are the rotation parameters of the intrinsic matrix, the translation parameters of the extrinsic matrix, and the distortion coefficients, respectively.

[0041] This application embodiment also provides a calibration device for a high-altitude visual measurement system, the device comprising: The first calibration unit is used to obtain the first calibration result, which is the calibration result of the binocular camera obtained by taking images of the ground target with the binocular camera when the target is placed on the ground; the field of view space when the binocular camera takes images of the ground target is recorded as the ground field of view space; The low-altitude and high-altitude imaging unit is used to acquire scene images of the ground-level field of view using a binocular camera after the first calibration is completed, and these images are recorded as the first scene images. The binocular camera is kept stationary and does not rotate horizontally, while the camera's pitch angle is increased so that the binocular camera's shooting range is aligned with a predetermined high-altitude area. The images captured in alignment with the predetermined high-altitude area are recorded as the second scene images. The field of view of the binocular camera when shooting the high-altitude area is recorded as the high-altitude field of view space. The bottom area of ​​the high-altitude field of view space overlaps with the top area of ​​the ground-level field of view space, and the overlapping area contains feature objects. Angle change determination unit is used to determine the pitch angle change value of the binocular camera from shooting the first scene image to shooting the second scene image based on the difference between the pixel data of the feature points of the feature object in the first scene image and the pixel data in the second scene image. The second calibration unit is used to determine the second calibration result based on the first calibration result and the pitch angle change value. The second calibration result represents the calibration parameters when the binocular camera is aimed at a high-altitude area for shooting.

[0042] Specific limitations regarding the calibration device for the high-altitude visual measurement system can be found in the limitations of the calibration method for the high-altitude visual measurement system described above, and will not be repeated here. Each unit in the aforementioned high-altitude visual measurement system calibration device can be implemented entirely or partially through software, hardware, or a combination thereof. These units can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each unit.

[0043] This application also provides a computer-readable storage medium storing computer instructions, which, when executed by a processor, implement the steps of the aforementioned high-altitude vision measurement system calibration method. The computer-readable storage medium may be a read-only memory (ROM), random access memory (RAM), compact disc read-only memory (CD-ROM), magnetic tape, floppy disk, or optical data storage node, etc.

[0044] This invention is not limited to the specific embodiments described above. Any modifications made by those skilled in the art based on the above concept without creative effort are within the scope of protection of this invention.

Claims

1. A calibration method for an aerial vision measurement system, characterized in that, Includes the following steps: Obtain the first calibration result, which is the calibration result of the binocular camera obtained by taking images of the ground target with the binocular camera when the target is placed on the ground; the field of view space when the binocular camera takes images of the ground target is recorded as the ground field of view space; After the first calibration is completed, use a binocular camera to collect scene images of the ground field of view, which are recorded as the first scene image; keep the binocular camera from translating or rotating horizontally, and increase the camera's pitch angle so that the binocular camera's shooting range is aligned with the predetermined high-altitude area, and record the image taken aligned with the predetermined high-altitude area as the second scene image. The field of view when the binocular camera captures a high-altitude area is denoted as the high-altitude field of view space. The bottom area of ​​the high-altitude field of view space overlaps with the top area of ​​the ground field of view space, and the overlapping area contains a feature object. Based on the difference between the pixel data of the feature points of the feature object in the first scene image and the pixel data in the second scene image, the pitch angle change value of the binocular camera from capturing the first scene image to capturing the second scene image is determined. Based on the first calibration result and the pitch angle change value, a second calibration result is determined. The second calibration result represents the calibration parameters when the binocular camera is aimed at a high-altitude area for shooting. The step of determining the second calibration result based on the first calibration result and the pitch angle change value includes: replacing the pitch angle in the extrinsic parameter matrix represented by the first parameter in the first calibration result with the sum of the pitch angle in the first calibration result and the pitch angle change value, wherein the first parameter includes the rotation angle between the world coordinate system and the camera coordinate system, and the rotation angle includes the rotation angle around the three coordinate axes as the rotation center; obtaining the updated extrinsic parameter matrix; and using the intrinsic parameter matrix in the first calibration result and the updated extrinsic parameter matrix as the second calibration result.

2. The calibration method for a high-altitude visual measurement system according to claim 1, characterized in that, The method for obtaining the first calibration result includes: Based on the four coordinate values ​​of the target point in the world coordinate system, camera coordinate system, image coordinate system, and pixel coordinate system, a first model representing the transformation relationship from the pixel coordinate system to the world coordinate system is obtained. The first model includes the camera's intrinsic parameter matrix and extrinsic parameter matrix, and the extrinsic parameter matrix includes a rotation matrix and a translation matrix. The camera is calibrated using the acquired target images to obtain the intrinsic and extrinsic parameter matrices. A second model is obtained to represent the relationship between the extrinsic parameter matrix parameters and the first parameter. The value of the first parameter is determined using the determined extrinsic parameter matrix parameters. The second model includes the first relationship between the rotation matrix and the rotation angle, as well as the second relationship between the coordinates of any point in space in the camera coordinate system and the coordinates in the world coordinate system, and the rotation matrix and the translation matrix.

3. The calibration method for a high-altitude visual measurement system according to claim 2, characterized in that, The first model, which obtains the transformation relationship from the pixel coordinate system to the world coordinate system based on the four coordinate values ​​of the target point in the world coordinate system, camera coordinate system, image coordinate system, and pixel coordinate system, includes: Let the matrix be... ;in, These are the target's coordinates in the pixel coordinate system; K is the target's coordinates in the world coordinate system; R is the intrinsic parameter matrix, T is the rotation matrix, and T is the translation matrix.

4. The calibration method for a high-altitude visual measurement system according to claim 2, characterized in that, The second model representing the relationship between the extrinsic parameter matrix parameters and the first parameter is: Camera rotation matrix for: , This represents the value of the element in the i-th row and j-th column of the matrix. It is related to the trigonometric function value of the rotation angle between the world coordinate system and the corresponding camera coordinate system; Camera translation matrix for: ;in, The coordinates of the target point in the world coordinate system. The coordinates of the target point in the camera coordinate system.

5. The calibration method for a high-altitude visual measurement system according to claim 1, characterized in that, The step of determining the pitch angle change value of the binocular camera from capturing the first scene image to capturing the second scene image based on the difference between the pixel data of the feature points of the feature object in the first scene image and the pixel data in the second scene image includes: The pixel position of the feature point of the feature object in the first scene image is recorded as the first pixel position. Obtain the pixel position of the feature point in the second scene image, and denote it as the second pixel position; The pitch angle change value is determined based on the positions of the first and second pixels.

6. The calibration method for a high-altitude visual measurement system according to claim 2, characterized in that, The process of calibrating the camera using the acquired target images to obtain intrinsic and extrinsic parameter matrices includes: After obtaining the coordinates of the target point in the world coordinate system and the pixel coordinate system, the coordinates in the world coordinate system and the pixel coordinate system are normalized to eliminate the difference in dimensions. The normalization method for the pixel coordinates of the target point includes the following steps: Get the pixel coordinates of all target points ( ), i = 1, 2, ..., n, where n is the number of target points; Calculate the centroid coordinates of all points, and use the average position of all points ( )express; Subtract the average position from the pixel coordinates of all target points so that the centroid of all points moves to the origin after each point is translated, thus obtaining the updated pixel coordinates of the target points. ); Calculate the average Euclidean distance from all points to the origin after translation. ; Calculate scaling factor , When the coordinates of a point are two-dimensional coordinates When the coordinates of a point are three-dimensional coordinates ; After translation, scale all points to obtain normalized coordinates. ), ; The normalization method for the world coordinate system of the target point is the same as the normalization method for the pixel coordinates of the target point described above.

7. The calibration method for a high-altitude visual measurement system according to claim 2, characterized in that, The process of calibrating the camera using the acquired target images to obtain intrinsic and extrinsic parameter matrices includes: Based on the collected target images, the pixel coordinates of multiple target points are obtained. Combined with the world coordinates of the target points and the first model, the H matrix is ​​solved. For the H matrix, the RQ decomposition is used to decompose the H matrix to obtain the intrinsic parameter matrix and the extrinsic parameter matrix; The optimal solutions for the intrinsic and extrinsic parameters are obtained by optimizing the matrix parameters using the LM algorithm.

8. A calibration device for a high-altitude visual measurement system, characterized in that, include: The first calibration unit is used to obtain the first calibration result, which is the calibration result of the binocular camera obtained by taking images of the ground target with the binocular camera when the target is placed on the ground; the field of view space when the binocular camera takes images of the ground target is recorded as the ground field of view space; The low-altitude and high-altitude shooting unit is used to collect scene images of the ground field of view space using a binocular camera after the first calibration is completed, and these images are recorded as the first scene images. The binocular camera is kept without translation or horizontal rotation, and the camera's pitch angle is increased so that the binocular camera's shooting range is aligned with the predetermined high-altitude area. The images captured aligned with the predetermined high-altitude area are recorded as the second scene images. The field of view when the binocular camera captures a high-altitude area is denoted as the high-altitude field of view space. The bottom area of ​​the high-altitude field of view space overlaps with the top area of ​​the ground field of view space, and the overlapping area contains a feature object. Angle change determination unit is used to determine the pitch angle change value of the binocular camera from shooting the first scene image to shooting the second scene image based on the difference between the pixel data of the feature points of the feature object in the first scene image and the pixel data in the second scene image. The second calibration unit is used to determine a second calibration result based on the first calibration result and the pitch angle change value. The second calibration result represents the calibration parameters when the binocular camera is aimed at a high-altitude area for shooting. The determination of the second calibration result based on the first calibration result and the pitch angle change value includes: replacing the pitch angle in the extrinsic parameter matrix represented by the first parameter in the first calibration result with the sum of the pitch angle in the first calibration result and the pitch angle change value. The first parameter includes the rotation angle between the world coordinate system and the camera coordinate system. The rotation angle includes the rotation angle around the three coordinate axes as the rotation center; obtaining the updated extrinsic parameter matrix; and using the intrinsic parameter matrix in the first calibration result and the updated extrinsic parameter matrix as the second calibration result.

9. A computer-readable storage medium storing computer instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the steps of the high-altitude visual measurement system calibration method as described in any one of claims 1-7.

Images