A dual 3D camera pair pose calibration method

Abstract

The application discloses a dual 3D camera pair shooting attitude calibration method, comprising the following steps: setting a calibration block, a plurality of holes vertically penetrating through the upper and lower surfaces of the calibration block are formed on the calibration block; establishing a calibration block coordinate system; simultaneously shooting the upper and lower surfaces of the calibration block through the first and second 3D cameras, and acquiring the first and second 3D camera coordinate system point clouds of the upper and lower surfaces of the calibration block; selecting a plurality of holes, and recording the coordinates of the selected holes on the upper and lower surfaces of the calibration block; according to the first and second 3D camera coordinate system point clouds, the coordinates of the mass centers of the selected holes on the upper and lower surfaces of the calibration block in the first and second 3D camera coordinate systems are respectively calculated; and according to the coordinates of the mass centers of the selected holes on the upper and lower surfaces of the calibration block in the first and second 3D camera coordinate systems and the coordinates of the selected holes on the upper and lower surfaces of the calibration block, the rigid transformation matrix of the first 3D camera attitude to the second 3D camera attitude, the second 3D camera attitude to the first 3D camera attitude or the first and second 3D camera attitudes to the calibration block attitude is obtained.

Description

Technical Field

[0001] This invention relates to the field of industrial automation machine vision technology, and in particular to a method for attitude calibration of dual 3D cameras. Background Technology

[0002] In the field of industrial automation machine vision, it is often necessary to know the pose relationships between multiple 3D cameras to achieve point cloud stitching and measurement. Currently, commonly used camera pose calibration methods are based on feature points in a common field of view. For example, "pagoda calibration" determines feature points through three faces and performs multi-camera pose calibration based on common features or the physical size of calibration blocks. Similar methods include "dot calibration block calibration" and "checkerboard calibration block calibration." However, these calibration methods cannot achieve the function of through-beam calibration. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to address at least one deficiency of the related technologies mentioned in the background: existing calibration methods cannot achieve the function of through-beam calibration, and to provide a dual 3D camera through-beam attitude calibration method.

[0004] The technical solution adopted by this invention to solve its technical problem is: to construct a dual 3D camera through-beam attitude calibration method, including the following steps:

[0005] Setting steps: Set up a calibration block, which has multiple holes that penetrate vertically through the upper and lower surfaces;

[0006] Establishment steps: Establish a three-dimensional calibration block coordinate system;

[0007] Acquisition steps: Simultaneously take photos of the upper and lower surfaces of the calibration block using a first 3D camera and a second 3D camera to obtain the point cloud of the upper surface of the calibration block in the first 3D camera coordinate system and the point cloud of the lower surface in the second 3D camera coordinate system.

[0008] Selection steps: Select multiple holes and record the coordinates of the selected holes on the upper and lower surfaces of the calibration block;

[0009] First calculation step: Calculate the coordinates of the centroid of the hole selected on the upper surface of the calibration block in the first 3D camera coordinate system based on the point cloud of the first 3D camera coordinate system;

[0010] Second calculation step: Calculate the coordinates of the centroid of the hole selected on the lower surface of the calibration block in the second 3D camera coordinate system based on the point cloud of the second 3D camera coordinate system;

[0011] Calibration steps: Based on the coordinates of the centroid of the hole selected on the upper surface of the calibration block in the first 3D camera coordinate system, the coordinates of the centroid of the hole selected on the lower surface of the calibration block in the second 3D camera coordinate system, and the coordinates of the selected hole on the upper and lower surfaces of the calibration block, obtain the rigid body transformation matrix from the first 3D camera pose to the second 3D camera pose, from the second 3D camera pose to the first 3D camera pose, or from the first 3D camera and the second 3D camera pose to the pose of the calibration block.

[0012] Preferably, in the dual 3D camera attitude calibration method of the present invention, the upper and lower surfaces of the calibration block are flat and parallel to each other.

[0013] Preferably, in the dual 3D camera through-beam attitude calibration method of the present invention, the establishment step includes:

[0014] A three-dimensional coordinate system is established using any of the holes on the upper or lower surface of the calibration block as the origin, thus obtaining the calibration block coordinate system.

[0015] Preferably, in the dual 3D camera through-beam attitude calibration method of the present invention, the selection step includes:

[0016] Select multiple holes in the coordinate system of the calibration block and record the coordinates of the selected holes on the upper and lower surfaces of the calibration block; or,

[0017] Multiple holes are selected in the first 3D camera coordinate system and the second 3D camera coordinate system, and the coordinates of the selected holes on the upper and lower surfaces of the calibration block are recorded.

[0018] Preferably, in the dual 3D camera through-beam attitude calibration method of the present invention, the positions of the plurality of holes selected in the first 3D camera coordinate system and the second 3D camera coordinate system are the same or different on the upper and lower surfaces of the calibration block.

[0019] Preferably, in the dual 3D camera through-beam attitude calibration method of the present invention, the first calculation step includes:

[0020] The plane equation of the upper surface of the calibration block is fitted by the point cloud of the first 3D camera coordinate system;

[0021] Calculate the distance from the selected hole and the points of the point cloud around it on the upper surface of the calibration block to the plane in the coordinate system of the calibration block to obtain a 2D distance map;

[0022] The region where the selected hole on the upper surface of the calibration block is located is extracted from the 2D distance map by using the distance threshold from the point to the plane, and the XY coordinates of the centroid of the selected hole on the upper surface of the calibration block in the first 3D camera coordinate system are determined.

[0023] Substituting the XY coordinates into the plane equation, we obtain the Z coordinate of the centroid of the hole selected on the upper surface of the calibration block in the first 3D camera coordinate system.

[0024] Finally, the XYZ coordinates of the centroid of the hole selected on the upper surface of the calibration block in the first 3D camera coordinate system are obtained;

[0025] The second calculation step includes:

[0026] The plane equation of the lower surface of the calibration block is fitted by the point cloud of the second 3D camera coordinate system;

[0027] Calculate the distance from the selected hole and the points of the point cloud around it on the lower surface of the calibration block in the coordinate system of the calibration block to the plane to obtain a 2D distance map;

[0028] The region containing the hole selected on the lower surface of the calibration block is extracted from the 2D distance map by using the distance threshold from the point to the plane, and the XY coordinates of the centroid of the hole selected on the lower surface of the calibration block in the second 3D camera coordinate system are determined.

[0029] Substituting the XY coordinates into the plane equation, we obtain the Z coordinate of the centroid of the hole selected on the lower surface of the calibration block in the second 3D camera coordinate system.

[0030] Finally, the XYZ coordinates of the centroid of the hole selected on the lower surface of the calibration block in the second 3D camera coordinate system are obtained.

[0031] Preferably, in the dual 3D camera through-beam attitude calibration method of the present invention, the step of fitting the plane equation of the upper surface of the calibration block using the point cloud of the first 3D camera coordinate system includes:

[0032] First, the point cloud of the selected plane where the hole is located is extracted using the point cloud of the first 3D camera coordinate system. Then, the plane equation of the upper surface of the calibration block is fitted using a random consistency sampling algorithm.

[0033] The step of fitting the plane equation of the lower surface of the calibration block using the point cloud of the second 3D camera coordinate system includes:

[0034] First, the point cloud of the selected plane containing the hole is extracted using the point cloud of the second 3D camera coordinate system. Then, the plane equation of the lower surface of the calibration block is fitted using a random consistency sampling algorithm.

[0035] Preferably, in the dual 3D camera attitude calibration method of the present invention, the least squares method is used to calculate the rigid body transformation matrix in the calibration step.

[0036] Preferably, in the dual 3D camera through-beam attitude calibration method of the present invention, the calibration step employs a random consistency sampling algorithm to calculate the rigid body transformation matrix, including:

[0037] S1: Extract the coordinates of the centroid of the hole selected on the upper surface of the calibration block in the first 3D camera coordinate system, the coordinates of the centroid of the hole selected on the lower surface of the calibration block in the second 3D camera coordinate system, and the coordinates of the selected hole on the upper and lower surfaces of the calibration block, and calculate a rigid body transformation matrix.

[0038] S2: Transform the centroid of the hole in different coordinate systems to the first 3D camera coordinate system, the second 3D camera coordinate system, or the calibration block coordinate system;

[0039] S3: If the centroids of the holes in different coordinate systems are uniformly transformed to the coordinate system of the calibration block, it is determined whether the distance between the transformed centroid coordinates of the holes and the centroid coordinates of the holes in the coordinate system of the calibration block is less than a distance threshold. If yes, it is considered that the centroids of the holes overlap; if no, it is considered that the centroids of the holes do not overlap.

[0040] S4: Repeat steps S1 to S3 multiple times to obtain multiple rigid body transformation matrices, and select the rigid body transformation matrix with the most overlap of the centroids of the holes as the final result.

[0041] Preferably, in the dual 3D camera through-beam attitude calibration method of the present invention, the method further includes:

[0042] Stitching steps: Use the rigid body transformation matrix to stitch the point cloud of the first 3D camera coordinate system and the point cloud of the second 3D camera coordinate system to obtain the stitched point cloud of the calibration block.

[0043] By implementing this invention, the following beneficial effects are achieved:

[0044] This invention discloses a method for attitude calibration of dual 3D cameras in a through-beam shooting configuration. This method enables high-precision calibration of the attitude relationship between cameras in situations where dual 3D cameras are shooting from opposite directions without a common field of view, with an error level within 0.01 mm. Furthermore, the calibration block used in this method is simple to manufacture, low in cost, and convenient for on-site calibration. It has few application limitations, and suitable calibration blocks can be customized to meet the needs of various complex dual-camera attitude calibration scenarios. Additionally, the attitude constraints between the camera and the calibration block are low during calibration; they do not require parallelism between the camera and the calibration block. It is only necessary to ensure that the upper and lower surfaces of the calibration block image normally under both the first and second 3D cameras, respectively. Attached Figure Description

[0045] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings:

[0046] Figure 1 This is a flowchart illustrating the dual 3D camera attitude calibration method of the present invention.

[0047] Figure 2 This is a schematic diagram of the specific calculation step in the dual 3D camera attitude calibration method of the present invention;

[0048] Figure 3 This is a schematic diagram of the specific calculation step in the dual 3D camera attitude calibration method of the present invention;

[0049] Figure 4 This is a schematic diagram illustrating the specific process of calculating the rigid body transformation matrix using the random consistency sampling algorithm in the calibration step of the dual 3D camera attitude calibration method of the present invention.

[0050] Figure 5 This is a schematic diagram of the process of point cloud stitching using rigid body transformation matrix in this invention;

[0051] Figure 6 This is a schematic diagram of the upper or lower surface structure of the calibration block of the present invention;

[0052] Figure 7 This is a schematic cross-sectional view of the calibration block of the present invention;

[0053] Figure 8 This is a depth map of the upper surface obtained by the first 3D camera of this invention;

[0054] Figure 9 This is a depth map of the lower surface obtained by the second 3D camera of this invention;

[0055] Figure 10 This is the first 3D camera coordinate system point cloud on the upper surface of the calibration block of this invention;

[0056] Figure 11This is the second 3D camera coordinate system point cloud on the lower surface of the calibration block of this invention;

[0057] Figure 12 This invention uses a rigid body transformation matrix to stitch together point clouds to obtain the point cloud of the calibration block. Detailed Implementation

[0058] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0059] It should be noted that the flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0060] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0061] In the field of industrial automation machine vision, multi-camera pose calibration is widely used, for example, in reverse engineering and workpiece dimension measurement. These applications require extremely high calibration accuracy. Furthermore, the application environments are complex and diverse, including confined spaces and mechanical interference, thus necessitating a calibration process that is as simple and easy to implement as possible. Therefore, such as Figure 1 As shown, one embodiment of the present invention discloses a method for attitude calibration of dual 3D cameras, comprising the following steps:

[0062] Setup steps: Set up a calibration block, which has multiple holes that penetrate vertically through the upper and lower surfaces.

[0063] Preferably, the upper and lower surfaces of the calibration block are flat and parallel to each other, and the holes are evenly spaced. Furthermore, the diameter of the holes and the spacing between the holes are proportional to the range and field of view of the 3D camera.

[0064] Specifically, making such Figure 6 and Figure 7The calibration block shown consists of a steel sheet of known thickness as its main body. The upper and lower surfaces of the steel sheet are parallel to each other and ensure flatness. Holes are drilled at equal intervals on the steel sheet, penetrating and perpendicular to the upper and lower surfaces. The diameter of the holes and the distance between them can be adjusted according to the range and field of view of the 3D camera used in the actual project. Generally, the larger the range and field of view, the larger the hole diameter and the greater the distance between the holes. Typically, the hole diameter is about 30 times the resolution of the 3D camera, for example, 25 to 35 times. The distance between the holes is generally about twice the hole diameter, preferably twice. In some other embodiments, the calibration block can be a sheet made of other metal or plastic materials.

[0065] Establishment steps: Establish a three-dimensional coordinate system for the calibration block. This includes: using any hole on the upper or lower surface of the calibration block as the origin, establishing a three-dimensional coordinate system to obtain the calibration block coordinate system.

[0066] Specifically, a three-dimensional coordinate system for the calibration block is artificially established based on the actual dimensions of the calibration block. Figure 6 Taking the calibration block shown as an example, the distance between the holes is 2 mm. If we establish a left-handed coordinate system with the hole at the top left corner of the upper surface of the calibration block as the origin, the XYZ coordinates of the origin are (0, 0, 0). The coordinates of the first hole to the right of the origin are (2, 0, 0), the coordinates of the second hole to the right of the origin are (4, 0, 0), and so on. The coordinates of the first point below the origin are (0, 2, 0), the coordinates of the second point below the origin are (0, 4, 0), and so on. Following this principle, the coordinates of the point in the 3rd column, 2nd row to the right of the origin are (6, 4, 0). Note that the Z-value of all holes on the upper surface of the calibration block is 0. Since the thickness of the calibration block is 2 mm, the Z-value of all holes on the lower surface of the calibration block is -2. For example, the coordinates of the hole corresponding to the origin on the lower surface of the calibration block are (0, 0, -2), and the coordinates of the point in the 3rd column, 2nd row on the lower surface are (6, 4, -2).

[0067] Acquisition steps: Simultaneously, the upper and lower surfaces of the calibration block are photographed using a first 3D camera and a second 3D camera to obtain the following results: Figure 8 The depth map of the upper surface obtained by the first 3D camera shown is as follows: Figure 9 The second 3D camera captured a depth map of the lower surface, as shown, and obtained the following: Figure 10 The first 3D camera coordinate system point cloud on the upper surface of the calibration block shown is as follows: Figure 11 The second 3D camera coordinate system point cloud is shown on the lower surface of the calibration block.

[0068] Selection steps: Select multiple holes and record the coordinates of the selected holes on the upper and lower surfaces of the calibration block. These recorded coordinates are in the coordinate system of the calibration block. Preferably, at least 4 holes are selected; the more holes selected, the higher the accuracy.

[0069] In some embodiments, the selection step includes:

[0070] Select multiple holes in the calibration block coordinate system and record the coordinates of the selected holes on the upper and lower surfaces of the calibration block. Selecting multiple holes in the calibration block coordinate system ensures that the same hole is located on both the upper and lower surfaces, which facilitates subsequent calculations.

[0071] In some embodiments, the selection step includes:

[0072] Multiple holes are selected in both the first and second 3D camera coordinate systems, and the coordinates of the selected holes on the upper and lower surfaces of the calibration block are recorded. Preferably, for ease of subsequent calculation, the multiple holes selected in the first and second 3D camera coordinate systems may be in the same or different positions on the upper and lower surfaces of the calibration block.

[0073] First calculation step: Based on the point cloud of the first 3D camera coordinate system, calculate the coordinates of the centroid of the selected hole on the upper surface of the calibration block in the first 3D camera coordinate system.

[0074] Specifically, such as Figure 2 As shown, the first calculation step includes:

[0075] The plane equation of the upper surface of the calibration block is fitted by the point cloud of the first 3D camera coordinate system. Specifically, the point cloud of the selected hole plane (i.e., the point cloud imaged on the upper surface of the calibration block) is first extracted from the point cloud of the first 3D camera coordinate system, and then the plane equation of the upper surface of the calibration block is fitted by the random consistency sampling algorithm.

[0076] Calculate the distances from the selected holes and surrounding point clouds on the upper surface of the calibration block to the plane in the calibration block coordinate system to obtain a 2D distance map;

[0077] Since the holes are through, the area where the selected holes are located on the upper surface of the calibration block is extracted from the 2D distance map by using the distance threshold from the point to the plane, and the XY coordinates of the centroid of the selected holes on the upper surface of the calibration block in the first 3D camera coordinate system are determined.

[0078] Substituting the XY coordinates into the plane equation, we obtain the Z coordinate of the centroid of the selected hole on the upper surface of the calibration block in the first 3D camera coordinate system.

[0079] Finally, the XYZ coordinates of the centroid of the selected hole on the upper surface of the calibration block in the first 3D camera coordinate system are obtained, which are the coordinates of the centroid of the hole projected onto the plane on the upper surface of the calibration block.

[0080] The second calculation step: Based on the point cloud of the second 3D camera coordinate system, calculate the coordinates of the centroid of the selected hole on the lower surface of the calibration block in the second 3D camera coordinate system.

[0081] Specifically, such as Figure 3 As shown, the second calculation step includes:

[0082] The plane equation of the lower surface of the calibration block is fitted by the point cloud of the second 3D camera coordinate system. Specifically, the point cloud of the selected hole plane (i.e., the point cloud imaged on the lower surface of the calibration block) is first extracted from the point cloud of the second 3D camera coordinate system, and then the plane equation of the lower surface of the calibration block is fitted by the random consistency sampling algorithm.

[0083] Calculate the distances from the selected holes and surrounding point clouds on the lower surface of the calibration block to the plane in the calibration block coordinate system to obtain a 2D distance map;

[0084] Since the holes are through, the area where the selected holes are located on the lower surface of the calibration block is extracted from the 2D distance map by using the distance threshold from the point to the plane, and the XY coordinates of the centroid of the selected holes on the lower surface of the calibration block in the second 3D camera coordinate system are determined.

[0085] Substituting the XY coordinates into the plane equation, we obtain the Z coordinate of the centroid of the selected hole on the lower surface of the calibration block in the second 3D camera coordinate system.

[0086] Finally, the XYZ coordinates of the centroid of the hole selected on the lower surface of the calibration block in the second 3D camera coordinate system are obtained, which are the coordinates of the centroid of the hole projected onto the plane where the lower surface of the calibration block is located.

[0087] Calibration steps: Based on the coordinates of the centroid of the selected hole on the upper surface of the calibration block in the first 3D camera coordinate system, the coordinates of the centroid of the selected hole on the lower surface of the calibration block in the second 3D camera coordinate system, and the coordinates of the selected hole on the upper and lower surfaces of the calibration block, obtain the rigid body transformation matrix from the first 3D camera pose to the second 3D camera pose, from the second 3D camera pose to the first 3D camera pose, or from the first 3D camera and second 3D camera poses to the calibration block pose. Here, pose refers to the pose of the camera setup.

[0088] In some embodiments, the least squares method is used to calculate the rigid body transformation matrix in the calibration step.

[0089] In some embodiments, the robustness of the calibration process can be further improved (by reducing noise interference or eliminating erroneous centroid coordinates). Since the least squares method is used to calculate the rigid body transformation matrix, all aperture coordinates in different coordinate systems are used. Therefore, if imaging noise causes errors in the calculation of aperture centroid coordinates or human error in selecting incorrect apertures, the resulting rigid body transformation matrix is ​​invalid. To address this issue, a random consistency sampling algorithm can be used in the calibration step to calculate the rigid body transformation matrix, such as... Figure 4 As shown, it specifically includes:

[0090] S1: Extract the coordinates of the centroid of the selected hole on the upper surface of the calibration block in the first 3D camera coordinate system, the coordinates of the centroid of the selected hole on the lower surface of the calibration block in the second 3D camera coordinate system, and the coordinates of the selected hole on the upper and lower surfaces of the calibration block, and calculate a rigid body transformation matrix.

[0091] S2: Transform the centroid of the hole in different coordinate systems to the first 3D camera coordinate system, the second 3D camera coordinate system, or the calibration block coordinate system;

[0092] S3: If the centroids of holes in different coordinate systems are uniformly transformed to the calibration block coordinate system, determine whether the distance between the transformed centroid coordinates of holes and the centroid coordinates of holes in the calibration block coordinate system is less than the distance threshold. If yes, the centroids of holes are considered to overlap; otherwise, the centroids of holes are considered not to overlap.

[0093] S4: Repeat steps S1 to S3 multiple times to obtain multiple rigid body transformation matrices. Select the rigid body transformation matrix with the most overlap of the hole centroids as the final result. For example, repeat 100 times, or the number of repetitions should be more than 5 times the number of holes selected.

[0094] In this embodiment, as Figure 5 As shown, the method also includes:

[0095] Stitching steps: Use a rigid body transformation matrix to stitch the point clouds of the first 3D camera coordinate system and the second 3D camera coordinate system together to obtain the point cloud of the stitched calibration block, as shown below. Figure 12 As shown, by splicing point clouds, the thickness of the hole calibration block was measured to be 1.99971 mm. Compared with the actual thickness of 2 mm, the error is very small. Therefore, it has practical application value in workpiece thickness detection, reverse engineering and other fields.

[0096] By implementing this invention, the following beneficial effects are achieved:

[0097] This invention discloses a method for attitude calibration of dual 3D cameras in a through-beam shooting configuration. This method enables high-precision calibration of the attitude relationship between cameras in situations where dual 3D cameras are shooting from opposite directions without a common field of view, with an error level within 0.01 mm. Furthermore, the calibration block used in this method is simple to manufacture, low in cost, and convenient for on-site calibration. It has few application limitations, and suitable calibration blocks can be customized to meet the needs of various complex dual-camera attitude calibration scenarios. Additionally, the attitude constraints between the camera and the calibration block are low during calibration; they do not require parallelism between the camera and the calibration block. It is only necessary to ensure that the upper and lower surfaces of the calibration block image normally under both the first and second 3D cameras, respectively.

[0098] It is understood that the above embodiments only illustrate some implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that for those skilled in the art, without departing from the concept of the present invention, the above embodiments or technical features can be freely combined, and several modifications and improvements can be made. These all fall within the protection scope of the present invention. That is, the embodiments described "in some embodiments" can be freely combined with any of the embodiments above and below. Therefore, all equivalent transformations and modifications made within the scope of the claims of the present invention should fall within the scope of the claims of the present invention.

Claims

1. A dual 3D camera pair pose calibration method, characterized in that, The method comprises the following steps: a setting step of setting a calibration block with a plurality of holes vertically penetrating through upper and lower surfaces; a establishing step of establishing a three-dimensional calibration block coordinate system; an acquiring step of acquiring a first 3D camera coordinate system point cloud of the upper surface and a second 3D camera coordinate system point cloud of the lower surface of the calibration block by simultaneously shooting the upper and lower surfaces of the calibration block through a first 3D camera and a second 3D camera; a selecting step of selecting a plurality of holes and recording the coordinates of the selected holes on the upper and lower surfaces of the calibration block; a first calculating step of calculating the coordinates of the selected hole centers in the upper surface of the calibration block in the first 3D camera coordinate system according to the first 3D camera coordinate system point cloud; a second calculating step of calculating the coordinates of the selected hole centers in the lower surface of the calibration block in the second 3D camera coordinate system according to the second 3D camera coordinate system point cloud; a calibrating step of obtaining a rigid transformation matrix of the first 3D camera pose to the second 3D camera pose, the second 3D camera pose to the first 3D camera pose or the first 3D camera and the second 3D camera pose to the calibration block pose according to the coordinates of the selected hole centers in the upper surface of the calibration block in the first 3D camera coordinate system, the coordinates of the selected hole centers in the lower surface of the calibration block in the second 3D camera coordinate system and the coordinates of the selected holes on the upper and lower surfaces of the calibration block.

2. The dual 3D camera pair pose calibration method of claim 1, wherein, The upper and lower surfaces of the calibration block are flat and parallel to each other.

3. The dual 3D camera pair pose calibration method of claim 1, wherein, The establishing step comprises: establishing a three-dimensional coordinate system with any of the holes in the upper surface or the lower surface of the calibration block as a coordinate origin to obtain the calibration block coordinate system.

4. The dual 3D camera pair pose calibration method of claim 1, wherein, The selecting step comprises: selecting a plurality of holes in the calibration block coordinate system and recording the coordinates of the selected holes on the upper and lower surfaces of the calibration block; or selecting a plurality of holes in the first 3D camera coordinate system and the second 3D camera coordinate system respectively and recording the coordinates of the selected holes on the upper and lower surfaces of the calibration block.

5. The dual 3D camera pair pose calibration method according to claim 4, characterized in that, The positions of the selected holes in the upper and lower surfaces of the calibration block are the same or different in the first 3D camera coordinate system and the second 3D camera coordinate system.

6. The dual 3D camera pair pose calibration method of claim 1, wherein, The first calculating step comprises: fitting a plane equation of the upper surface of the calibration block through the first 3D camera coordinate system point cloud; calculating the distances from the points of the selected holes and their surrounding point clouds in the upper surface of the calibration block in the calibration block coordinate system to the plane to obtain a 2D distance map; extracting the regions where the selected holes are located in the upper surface of the calibration block from the 2D distance map through a distance threshold of the points to the plane and determining the XY coordinates of the selected hole centers in the upper surface of the calibration block in the first 3D camera coordinate system; substituting the XY coordinates into the plane equation to obtain the Z coordinates of the selected hole centers in the upper surface of the calibration block in the first 3D camera coordinate system; finally obtaining the XYZ coordinates of the selected hole centers in the upper surface of the calibration block in the first 3D camera coordinate system. The second calculation step comprises: fitting a plane equation of the lower surface of the calibration block through the point cloud of the second 3D camera coordinate system; calculating the distance from the selected hole and its surrounding point cloud on the lower surface of the calibration block to the plane in the calibration block coordinate system, obtaining a 2D distance map; extracting the area where the selected hole is located from the 2D distance map by setting a distance threshold, and determining the XY coordinates of the selected hole centroid on the lower surface of the calibration block in the second 3D camera coordinate system; substituting the XY coordinates into the plane equation to obtain the Z coordinate of the selected hole centroid on the lower surface of the calibration block in the second 3D camera coordinate system; finally obtaining the XYZ coordinates of the selected hole centroid on the lower surface of the calibration block in the second 3D camera coordinate system.

7. The dual 3D camera pair pose calibration method of claim 6, wherein, The plane equation fitting of the upper surface of the calibration block through the point cloud of the first 3D camera coordinate system comprises: firstly, extracting the point cloud of the plane where the selected hole is located through the point cloud of the first 3D camera coordinate system, and then fitting the plane equation of the upper surface of the calibration block through the random consistency sampling algorithm; The plane equation fitting of the lower surface of the calibration block through the point cloud of the second 3D camera coordinate system comprises: firstly, extracting the point cloud of the plane where the selected hole is located through the point cloud of the second 3D camera coordinate system, and then fitting the plane equation of the lower surface of the calibration block through the random consistency sampling algorithm.

8. The dual 3D camera pair pose calibration method of claim 1, wherein, The least square method is used for calculating the rigid transformation matrix in the calibration step.

9. The dual 3D camera pair pose calibration method of claim 1, wherein, The random consistency sampling algorithm is used for calculating the rigid transformation matrix in the calibration step, comprising: S1: respectively extracting a group of coordinates of the selected hole centroid on the upper surface of the calibration block in the first 3D camera coordinate system, coordinates of the selected hole centroid on the lower surface of the calibration block in the second 3D camera coordinate system, and coordinates of the selected hole on the upper and lower surfaces of the calibration block, and calculating a rigid transformation matrix; S2: transforming the hole centroids in different coordinate systems to the first 3D camera coordinate system, the second 3D camera coordinate system or the calibration block coordinate system; S3: if the hole centroids in different coordinate systems are transformed to the calibration block coordinate system, judging whether the distance between the converted hole centroid coordinates and the hole centroid coordinates in the calibration block coordinate system is less than the distance threshold, if yes, considering that the hole centroids overlap; if no, considering that the hole centroids do not overlap; S4: repeating the above steps S1 to S3 for multiple times to obtain multiple rigid transformation matrices, and selecting the rigid transformation matrix with the most overlapping hole centroids as the final result.

10. The dual 3D camera pair pose calibration method of claim 1, wherein, The method further comprises: splicing step: splicing the point cloud of the first 3D camera coordinate system and the point cloud of the second 3D camera coordinate system by using the rigid transformation matrix, obtaining the spliced point cloud of the calibration block.

Images