A Target and Matching Calibration Method for LiDAR and Camera Extrinsic Parameter Calibration

Abstract

The application discloses a laser radar and camera external parameter calibration target and a matching calibration method, and comprises the following steps: acquiring a frame of point cloud collected by a laser radar on a calibration target; acquiring a frame of camera image collected by a camera on the same position of the calibration target; extracting three planes of the calibration target in the point cloud, merging the point cloud of the same plane, and calculating the three-dimensional intersection line of the planes of the calibration target; extracting the gray edge of the calibration target in the camera image to obtain an edge image; projecting the three-dimensional intersection line onto the edge image to construct a constraint optimization problem; and iteratively solving the constraint optimization problem to obtain the laser radar and camera external parameters which minimize the distance between the three-dimensional intersection line and the gray edge. The method greatly improves the automation level of the laser radar and camera external parameter calibration process, improves the laser radar and camera external parameter calibration precision, and finally reduces the re-projection error of the point cloud feature to the image to within one pixel.

Description

Technical Field

[0001] This invention belongs to the technical field of sensor calibration technology and image processing technology, specifically involving a calibration target and matching calibration method for external parameters of lidar and camera. Background Technology

[0002] LiDAR and visible light cameras are complementary sensors. LiDAR can directly obtain 3D point cloud information by measuring the laser emission and reception time, but the point cloud information is sparse and does not contain texture information; visible light cameras cannot obtain 3D information of a scene, but they have rich texture information. LiDAR is currently widely used in the field of autonomous driving.

[0003] The existing external parameter calibration targets and corresponding calibration methods for some lidar and camera systems are as follows:

[0004] 1. Technology based on chessboard calibration board

[0005] Examples include CN117830438A (A joint calibration method for lidar and camera based on specific markers) and CN117934629A (An adaptive joint calibration method and apparatus for lidar and camera).

[0006] Main principle:

[0007] The camera's intrinsic parameters are obtained by constructing a homography matrix between the object and the image based on the checkerboard image acquired by the camera. The position of the calibration board is detected in the image using the grayscale features of the checkerboard. The position of the calibration board is also detected in the LiDAR point cloud using the geometric features of the checkerboard. The extrinsic parameters of the LiDAR and camera are calculated based on the correspondence between the calibration board in the image and the point cloud.

[0008] Method limitations:

[0009] Checkerboard calibration boards fail to provide robust geometric features for LiDAR. LiDAR becomes unreliable when detecting the edges of the calibration board, and the grayscale features of the checkerboard pattern do not aid in detection within point clouds. Existing improvements largely only alter the shape of the calibration board, offering limited enhancement to LiDAR detection performance and thus limiting the final calibration results.

[0010] 2. Techniques based on object features in natural scenes

[0011] Examples include CN117953082A (A method, system and electronic device for joint calibration of lidar and camera) and CN117808895A (A method and related device for targetless extrinsic parameter calibration of lidar and camera).

[0012] Main principle:

[0013] The system acquires initial images and point cloud data of the target building by simultaneously scanning the target building at multiple different distances from the camera and lidar. From the multiple initial images scanned at the same distance, the target image with the scan time being the target time is selected, and the image features of the target image are extracted. The point cloud features are extracted from the point cloud data. The image features and point cloud features are correlated, and the minimum feature distance between the correlated image features and point cloud features is calculated to obtain the optimized calibration extrinsic parameters.

[0014] Method limitations:

[0015] Challenges in Feature Extraction and Matching: In natural scenes, the shape, material, and ambient lighting conditions of object features are often complex, posing challenges to feature extraction and matching. The stability and repeatability of features may also be affected.

[0016] The complexity of scene structure: Natural scenes are often structurally complex, with objects of varying shapes, sizes, and distributions. This increases the difficulty of matching object features and may lead to biases in the calibration results.

[0017] Impact of environmental changes: The natural environment may change due to factors such as sunlight, shading, and weather. This can make it difficult to repeatedly detect previously calibrated feature points, thus affecting the stability and reliability of the calibration results.

[0018] Limitations on calibration accuracy: Even if feature points in natural scenes can be successfully extracted and matched, the localization accuracy may not reach the level of checkerboard calibration, which limits the improvement of overall calibration accuracy.

[0019] High computational complexity: Calibration methods based on natural scenes usually involve a large amount of image processing and feature detection calculations, resulting in high computational complexity and high requirements for system real-time performance and computing resources.

[0020] Currently, there is a lack of a simple calibration process and a high-precision calibration method for the extrinsic parameters of lidar and cameras. Summary of the Invention

[0021] To address the aforementioned technical problems, this invention provides a calibration target and a matching calibration method for external parameters of lidar and cameras.

[0022] The technical solution provided by this invention is as follows:

[0023] In a first aspect, the present invention provides a target and a matching calibration method for external parameter calibration of a lidar and camera, comprising the following steps:

[0024] Acquire a point cloud frame obtained by lidar calibration of the target;

[0025] Acquire a single frame of camera image captured by the camera at the same location as the calibration target;

[0026] Extract the three planes of the calibration target from the point cloud, merge the point clouds of the same plane, and calculate the three-dimensional intersection line of the calibration target plane;

[0027] Extract the grayscale edges of the calibrated target from the camera image to obtain the edge image;

[0028] Projecting the 3D intersection line onto the edge image, we construct a constrained optimization problem.

[0029] By iteratively solving the constrained optimization problem, the extrinsic parameters of the lidar and camera that minimize the distance between the 3D intersection line and the grayscale edge are obtained.

[0030] In one possible implementation, the calibration target is made of three planar plates with edges, the three planes being adjacent to each other and at a certain angle.

[0031] In one possible implementation, the method for merging point clouds on the same plane comprises the following steps: merging point clouds on the same plane by detecting the distance between the extracted planes and the plane normal vectors.

[0032] In one possible implementation, the method for calculating the three-dimensional intersection line of the calibration target plane includes the following steps: calculating the three-dimensional intersection line of the calibration target plane by fitting the plane equation of the target;

[0033] The equation of the plane is:

[0034] Ax + By + Cz + D = 0

[0035] in, To obtain the plane normal vector using a plane point cloud, the superscript T represents the transpose operator, and A, B, and C are the three components of the plane normal vector in spatial coordinates. D is obtained by substituting the average coordinates (x, y, z) of the plane point cloud into the equation.

[0036] In one possible implementation, the method for extracting the grayscale edge of the calibration target in the camera image to obtain the edge image is as follows: the camera image is grayscaled and Gaussian blurred, and then the Canny operator is used to extract the grayscale edge of the calibration target in the image.

[0037] In one possible implementation, the distance between the 3D intersection of the projection and the grayscale edge is calculated, and a constrained optimization problem based on the extrinsic parameters of the lidar and camera is constructed.

[0038] The objective function E of the constructed constrained optimization problem is:

[0039]

[0040] in, The normal vector represents the grayscale edge feature, where i represents the corresponding i-th data group, and the superscript T is the vector transpose operator; Let δT represent the initial values ​​of the extrinsic parameters of the lidar and camera, where C represents the camera and L represents the lidar; δT represents the change in the extrinsic parameters of the lidar and camera from the initial values ​​to the optimal solution, expressed as Lie algebra; Exp() represents the exponential transformation from Lie algebra to the rotation and translation matrix; f() represents the projection transformation from the 3D point cloud acquired by the lidar to the 2D image plane through the camera intrinsic parameter matrix. Represents the feature points of the lidar; q represents the three-dimensional measurement error of the lidar; i Represents grayscale edge feature points on an image; This indicates the pixel error of the image captured by the camera.

[0041] Furthermore, the method for iteratively solving constrained optimization problems is as follows:

[0042] Calculate the Jacobian matrix of the objective function E with respect to the parameter δT, and then update the parameter using the Gauss-Newton method in each iteration. By solving this constrained optimization problem, obtain the extrinsic parameters of the lidar and camera that minimize the distance between the 3D intersection line of the projection and the grayscale edge. The extrinsic parameters include matrices describing the relative rotation and translation between the camera and lidar.

[0043] Secondly, the present invention provides a laser radar and camera extrinsic parameter calibration target and a matching calibration device, comprising:

[0044] The acquisition module is used to acquire a point cloud frame collected by the lidar on the calibration target and a camera image frame collected by the camera at the same position on the calibration target.

[0045] The 3D intersection calculation module is used to extract the three planes of the calibration target from the point cloud, merge the point clouds of the same plane, and calculate the 3D intersection of the calibration target planes.

[0046] The edge image extraction module is used to extract the grayscale edges of the calibrated target in the camera image to obtain the edge image;

[0047] The construction module is used to project 3D intersection lines onto the edge image to construct a constrained optimization problem.

[0048] The solution module is used to iteratively solve the constrained optimization problem and obtain the extrinsic parameters of the LiDAR and camera that minimize the distance between the 3D intersection line and the grayscale edge.

[0049] Thirdly, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the lidar and camera extrinsic calibration target and the matching calibration method as described above.

[0050] Fourthly, the present invention provides a non-transitory computer-readable storage medium storing a computer program thereon, wherein the computer program, when executed by a processor, implements the aforementioned lidar and camera extrinsic parameter calibration target and matching calibration method.

[0051] The beneficial effects of this invention are as follows:

[0052] (1) The calibration target described in this invention is different from the checkerboard calibration board in that it has obvious color features, which can be easily extracted from the image; it also has obvious geometric features, which can be easily extracted from the lidar point cloud.

[0053] (2) The method described in this invention has a direct and concise correspondence between the features extracted from the image and the features extracted from the lidar point cloud, which is beneficial for constructing a constrained optimization problem to solve the lidar and camera extrinsic parameters.

[0054] (3) The method of this invention extracts the color features of the calibration target from the image and the geometric features of the calibration target from the LiDAR point cloud. Then, the two features are matched, and a constrained optimization problem regarding the extrinsic parameters of the LiDAR and camera is constructed. This problem is solved iteratively to obtain the accurate extrinsic parameters of the LiDAR and camera. This method greatly improves the automation level of the LiDAR and camera extrinsic parameter calibration process, enhances the calibration accuracy of the LiDAR and camera extrinsic parameters, and ultimately reduces the reprojection error from point cloud features to the image to within one pixel. Attached Figure Description

[0055] Figure 1 Two structural design diagrams for the calibration target;

[0056] Figure 2 This section describes the extraction of the 3D intersection line of the calibration target plane. The blue line represents the extracted 3D intersection line of the calibration target plane. Other colored lines represent the plane extraction of the point cloud, with different colors indicating that points belong to different planes.

[0057] Figure 3 Extracting grayscale edges of the calibration target from the camera image. The left image is the original image captured by the camera, and the right image is the final extracted grayscale edge of the calibration target;

[0058] Figure 4This involves constructing a constraint optimization problem by projecting the 3D intersection line of the point cloud plane onto the edge plane extracted from the image. The blue part represents the edge plane extracted from the image, the red part represents the result of projecting the 3D intersection line of the plane onto the image plane, and the green part represents the matching and distance between the intersection line of the image and the point cloud plane.

[0059] Figure 5 A schematic diagram of the structure of the external parameter calibration target and matching calibration device for the laser radar and camera provided for the invention;

[0060] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0061] The present invention will be further described below with reference to specific embodiments, but the content of the present invention is not limited thereto.

[0062] A laser radar and camera extrinsic calibration target and matching calibration method, comprising the following steps:

[0063] S100, acquire a frame of point cloud data collected by the lidar calibration target;

[0064] S200: Acquire a single frame of camera image captured by the camera at the same location as the calibration target;

[0065] S300 extracts three planes of the calibration target from the point cloud, merges the point clouds of the same plane, and calculates the three-dimensional intersection line of the calibration target plane;

[0066] S400 extracts the grayscale edges of the calibrated target in the camera image to obtain the edge image;

[0067] S500 projects the 3D intersection line onto the edge image to construct a constrained optimization problem.

[0068] S600 iteratively solves the constrained optimization problem to obtain the extrinsic parameters of the LiDAR and camera that minimize the distance between the 3D intersection line and the grayscale edge.

[0069] In one possible implementation, the calibration target is made of three planar plates with edges, the three planes being adjacent to each other and at a certain angle.

[0070] In one possible implementation, the method for merging point clouds on the same plane in S300 involves the following steps: merging point clouds on the same plane by detecting the distance between the extracted planes and the plane normal vectors.

[0071] In one possible implementation, the method for calculating the three-dimensional intersection line of the calibration target plane in S300 includes the following steps: calculating the three-dimensional intersection line of the calibration target plane using plane equations;

[0072] The equation of the plane is:

[0073] Ax + By + Cz + D = 0

[0074] in, To obtain the plane normal vector using a plane point cloud, the superscript T represents the transpose operator, and A, B, and C are the three components of the plane normal vector in spatial coordinates. D is obtained by substituting the average coordinates (x, y, z) of the plane point cloud into the equation.

[0075] In one possible implementation, in S400, the method for extracting the grayscale edge of the calibration target in the camera image to obtain the edge image is as follows: the camera image is grayscaled and Gaussian blurred, and then the Canny operator is used to extract the grayscale edge of the calibration target in the image.

[0076] In one possible implementation, in S500, the distance between the 3D intersection of the projection and the grayscale edge is calculated, and a constrained optimization problem based on the extrinsic parameters of the lidar and camera is constructed.

[0077] Furthermore, the objective function E of the constructed constrained optimization problem is:

[0078]

[0079] in, The normal vector represents the grayscale edge feature, where i represents the corresponding i-th data group, and the superscript T is the vector transpose operator; Let δT represent the initial values ​​of the extrinsic parameters of the lidar and camera, where C represents the camera and L represents the lidar; δT represents the change in the extrinsic parameters of the lidar and camera from the initial values ​​to the optimal solution, expressed as Lie algebra; Exp() represents the exponential transformation from Lie algebra to the rotation and translation matrix; f() represents the projection transformation from the 3D point cloud acquired by the lidar to the 2D image plane through the camera intrinsic parameter matrix. Represents the feature points of the lidar; q represents the three-dimensional measurement error of the lidar; i Represents grayscale edge feature points on an image; This indicates the pixel error of the image captured by the camera.

[0080] In one possible implementation, the method for iteratively solving the constrained optimization problem in S600 is as follows:

[0081] Calculate the Jacobian matrix of the objective function E with respect to the parameter δT, and then update the parameters using the Gauss-Newton method in each iteration. By solving this constrained optimization problem, obtain the extrinsic parameters of the lidar and camera that minimize the distance between the 3D intersection of the projection and the grayscale edge. The extrinsic parameters mainly include matrices describing the relative rotation and translation between the camera and the lidar.

[0082] The following will illustrate this further with more specific examples.

[0083] A laser radar and camera extrinsic calibration target and matching calibration method, the steps of which are as follows:

[0084] (1) Place the external parameter calibration target of the lidar and camera in the common field of view of the lidar and camera, at a distance of about 5 meters. Collect a point cloud frame of the calibration target through the lidar and a camera image frame of the same calibration target through the camera.

[0085] The calibration target is composed of three coupled planar plates, each with a distinctly different color (the specific color is not strictly required), and the exact shape of the three plates is also not strictly required. The three planar plates are at a certain angle, allowing them to be detected simultaneously by both the lidar and the camera. Figure 1 The diagram shows two specific design schematics for LiDAR and camera extrinsic calibration targets, differing only in color.

[0086] (2) The RANSAC algorithm is used to extract the target plane from the lidar point cloud, and the point clouds of the same plane are merged by detecting the distance between the extracted planes and the plane normal vector. The three-dimensional intersection line of the calibration target plane is calculated by plane equation, and the result is as follows. Figure 2 As shown.

[0087] (3) First, the camera image is converted to grayscale and then Gaussian blurred. Then, the Canny operator is used to extract the grayscale edges of the target in the image. The result is as follows: Figure 3 As shown. Because different planes correspond to different colors, the grayscale edge corresponds to the three-dimensional intersection of the planes.

[0088] (4) Project the 3D intersection line of the plane extracted in (2) onto the edge image extracted in (3). For the calibration target designed in this embodiment, the grayscale edge corresponds to the 3D intersection line of the plane. Therefore, calculate the distance between the projected 3D intersection line and the grayscale edge on the image plane, and construct a constrained optimization problem based on the extrinsic parameters of the LiDAR and camera. The construction of the constrained optimization problem is as follows: Figure 4 As shown.

[0089] The objective function E of the constructed constrained optimization problem is:

[0090]

[0091] in, The normal vector represents the grayscale edge feature, where i represents the corresponding i-th data group, and the superscript T is the vector transpose operator; Let δT represent the initial values ​​of the extrinsic parameters of the lidar and camera, where C represents the camera and L represents the lidar; δT represents the change in the extrinsic parameters of the lidar and camera from the initial values ​​to the optimal solution, expressed as Lie algebra; Exp() represents the exponential transformation from Lie algebra to the rotation and translation matrix; f() represents the projection transformation from the 3D point cloud acquired by the lidar to the 2D image plane through the camera intrinsic parameter matrix. Represents the feature points of the lidar; q represents the three-dimensional measurement error of the lidar; i Represents grayscale edge feature points on an image; This indicates the pixel error of the image captured by the camera.

[0092] (5) Iteratively solve the constraint optimization problem to obtain the extrinsic parameters of the lidar and camera that minimize the distance between the three-dimensional intersection line and the grayscale edge.

[0093] This optimization problem is solved iteratively using the Gauss-Newton method. The iterative method involves calculating the Jacobian matrix of the objective function E with respect to the parameter δT, and then updating the parameters using the Gauss-Newton formula in each iteration. By solving this constrained optimization problem, the extrinsic parameters of the LiDAR and camera that minimize the distance between the 3D intersection line of the projection and the grayscale edge are obtained. These extrinsic parameters mainly include matrices describing the relative rotation and translation between the camera and the LiDAR. To prevent obtaining local optima, multiple sets of data are collected and used simultaneously in the optimization process.

[0094] The following describes the extrinsic parameter calibration target and matching calibration device for lidar and camera provided by the present invention. The extrinsic parameter calibration target and matching calibration device described below can be referred to in correspondence with the extrinsic parameter calibration target and matching calibration method described above.

[0095] Figure 5 This is a schematic diagram of the structure of the lidar and camera extrinsic calibration target and matching calibration device provided in an embodiment of the present invention, as shown below. Figure 5 As shown, it includes: an acquisition module 51, a 3D intersection calculation module 52, an edge image extraction module 53, a construction module 54, and a solution module 55, wherein:

[0096] The acquisition module 51 is used to acquire a point cloud frame collected by the lidar on the calibration target and a camera image frame collected by the camera at the same position on the calibration target.

[0097] The three-dimensional intersection calculation module 52 is used to extract the three planes of the calibration target in the point cloud, merge the point clouds of the same plane, and calculate the three-dimensional intersection of the calibration target planes.

[0098] The edge image extraction module 53 is used to extract the grayscale edges of the calibration target in the camera image to obtain the edge image;

[0099] Construction module 54 is used to project the 3D intersection line onto the edge image to construct a constrained optimization problem;

[0100] The solver module 55 is used to iteratively solve the constrained optimization problem and obtain the extrinsic parameters of the lidar and camera that minimize the distance between the 3D intersection line and the grayscale edge.

[0101] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown, the electronic device may include: a processor 610, a communication interface 620, a memory 630, and a communication bus 640, wherein the processor 610, the communication interface 620, and the memory 630 communicate with each other through the communication bus 640. The processor 610 can call the logical instructions in the memory 630 to execute the external parameter calibration target and the corresponding calibration method for the lidar and camera.

[0102] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0103] On the other hand, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, is implemented to perform the lidar and camera extrinsic calibration target and the associated calibration method provided by the above methods.

[0104] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0105] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0106] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A calibration method matched with a laser radar-camera external parameter calibration target, characterized in that, The method comprises the following steps: acquire a frame of point cloud collected by a laser radar on a calibration target; the calibration target is made of three planar panels with edges, and the three planes are adjacent and at a certain angle; acquire a frame of camera image collected by a camera on the same position of the calibration target; extract the three planes of the calibration target in the point cloud, merge the point cloud of the same plane, and calculate the three-dimensional intersection line of the planes of the calibration target; extract the gray edges of the calibration target in the camera image to obtain an edge image; project the three-dimensional intersection line onto the edge image to construct a constraint optimization problem; iteratively solve the constraint optimization problem to obtain the laser radar and camera extrinsic parameters when the three-dimensional intersection line is closest to the gray edges.

2. The calibration method matched with the laser radar-camera external parameter calibration target according to claim 1, characterized in that, The method for merging the point cloud of the same plane comprises the following steps: merge the point cloud of the same plane by detecting the distance between the extracted planes and the plane normal vector.

3. The calibration method matched with the laser radar-camera external parameter calibration target according to claim 1, characterized in that, The method for calculating the three-dimensional intersection line of the planes of the calibration target comprises the following steps: calculate the three-dimensional intersection line of the planes of the calibration target by fitting the plane equation of the target; the plane equation is: wherein, is the plane normal vector solved by the plane point cloud, the upper subscript T represents the transpose operator, A, B, and C are three components of the plane normal vector in the space coordinates, and the average coordinates (x, y, z) of the plane point cloud are brought into the equation to solve .

4. The calibration method with the LIDAR-camera extrinsic calibration target according to claim 1, characterized in that, The method for extracting the gray edges of the calibration target in the camera image to obtain an edge image comprises the following steps: perform gray processing and Gaussian blur on the camera image, and then use a Canny operator to extract the gray edges of the calibration target in the image.

5. The calibration method with the LIDAR-camera extrinsic calibration target according to claim 1, wherein, calculate the distance between the projected three-dimensional intersection line and the gray edges, and construct a constraint optimization problem based on the laser radar and camera extrinsic parameters; Objective function of the constructed constrained optimization problem E is: wherein, a normal vector representing a gray scale edge feature, i a corresponding first i group of data, the upper index T is the transposition operator of the vector; represents the initial value of the laser radar and camera extrinsic parameters, C represents the camera, L represents the laser radar; represents the change of the laser radar and camera extrinsic parameters from the initial value to the optimal solution, expressed by Lie algebra; represents the exponential transformation from Lie algebra to rotation and translation matrix; represents the projection transformation from the three-dimensional point cloud collected by the laser radar to the two-dimensional image plane through the camera intrinsic parameter matrix; represents the laser radar feature point; represents the three-dimensional measurement error of the laser radar; represents the gray scale edge feature point on the image; represents the pixel error of the image collected by the camera.

6. The calibration method matched with the laser radar-camera external parameter calibration target according to claim 5, characterized in that, The method for iteratively solving the constraint optimization problem comprises the following steps: Computing the objective function E Regarding the parameters of the Jacobian matrix, then in each iteration, the parameters are updated using the Gauss-Newton formula; by solving this constrained optimization problem, the laser radar and camera extrinsic parameters that minimize the distance between the projected three-dimensional intersection line and the gray edge are obtained, and the extrinsic parameters include the matrix describing the relative rotation and translation between the camera and the laser radar.

7. A calibration device matched with a laser radar-camera external parameter calibration target, characterized in that, comprise: an acquisition module configured to acquire a frame of point cloud collected by a laser radar on a calibration target and a frame of camera image collected by a camera on the same position of the calibration target; the calibration target is made of three planar panels with edges, and the three planes are adjacent and at a certain angle; a three-dimensional intersection line calculation module configured to extract the three planes of the calibration target in the point cloud, merge the point cloud of the same plane, and calculate the three-dimensional intersection line of the planes of the calibration target; an edge image extraction module configured to extract the gray edges of the calibration target in the camera image to obtain an edge image; a construction module configured to project the three-dimensional intersection line onto the edge image to construct a constraint optimization problem; a solving module configured to iteratively solve the constraint optimization problem to obtain the laser radar and camera extrinsic parameters when the three-dimensional intersection line is closest to the gray edges.

8. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the calibration method matched with the laser radar-camera extrinsic parameter calibration target according to any one of claims 1 to 6 when executing the program. 9.A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program implements the calibration method matched with the laser radar-camera extrinsic parameter calibration target according to any one of claims 1 to 6 when executed by the processor.

Images