A camera calibration method based on area array radar point cloud data and a storage medium

Abstract

The camera calibration method based on the planar array radar point cloud data and the storage medium, the camera calibration method comprises the following steps: a. the point cloud data acquired by the point cloud camera is segmented according to the characteristics of space, set and texture, then the segmented point cloud data is calculated by plane equation, and the plane equation (AX+BY+CZ+D=0) of the ground is acquired; b. the normal vector of the ground is (A, B, C); c. the rotation axis between two planes and the rotation axis included angle theta are calculated. The camera calibration method, the point cloud camera shoots the environment to acquire point cloud data, then the point cloud data is analyzed to acquire the plane equation of the ground, and the rotation axis and the rotation axis included angle theta of the relative world coordinate system data deviation are calculated, finally the calculated rotation axis and the rotation axis included angle theta are used as the position deviation compensation when the on-site environment is shot, so that the adjustment position deviation of the point cloud camera needs to be manually adjusted multiple times, debugging is convenient, and the universality is high.

Description

Technical Field

[0001] This invention belongs to the field of object visual recognition technology, specifically relating to a camera calibration method and storage medium based on area array radar point cloud data. Background Technology

[0002] In recent years, with the rise of smart factories and unmanned warehouses, as well as the transformation and upgrading of traditional manufacturing, laser-guided forklifts, as an important type of AGV (Automated Guided Vehicle) equipment, have been widely used in numerous automated production lines. However, as AGVs become increasingly intelligent, their reliance on vision cameras is gradually increasing. In practical applications, vision cameras enable visual navigation and 3D obstacle avoidance, as well as object recognition.

[0003] In real-world applications, point cloud cameras are not always installed perfectly and fixed in their positions as ideally as they might seem. This is usually due to machining errors in the mechanical holes, slight deformations of sheet metal parts, or human-induced angular deviations. These deviations cause the data acquired by the point cloud camera to shift angularly, requiring multiple manual adjustments to the camera's position, making the process quite cumbersome. Summary of the Invention

[0004] The purpose of this invention is to overcome the problem of cumbersome operation caused by the need to adjust the position deviation multiple times when installing a point cloud camera on an AGV, and to provide a camera calibration method and storage medium based on area array radar point cloud data that automatically compensates for position deviation.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A camera calibration method based on area array radar point cloud data, characterized by the following steps:

[0007] a. The point cloud data acquired by the point cloud camera is segmented according to the spatial, ensemble and texture features. Then, the segmented point cloud data is used to verify the plane equation to obtain the plane equation of the ground (AX+BY+CZ+D=0).

[0008] b. Obtain the normal vector of the ground based on the plane equation of the ground (AX+BY+CZ+D=0). The coordinates are (A, B, C).

[0009] c. Use the ground's normal vector By performing a cross product with the Z-axis unit vector (0, 0, 1) of the world coordinate system, we can obtain the rotation axis between the two planes and the angle θ between the rotation axes.

[0010] Compared with existing technologies, the camera calibration method based on area array radar point cloud data of the present invention uses a point cloud camera to capture point cloud data of the environment, then uses the point cloud data to analyze and obtain the plane equation of the ground, and calculates the rotation axis and the included angle θ of the rotation axis relative to the world coordinate system data. Finally, the calculated rotation axis and the included angle θ are used as position deviation compensation when capturing images of the on-site environment. This effectively eliminates the need for manual adjustment of the point cloud camera position deviation, making debugging convenient. Moreover, it is applicable to AGV vehicles with different point cloud camera mountings, and has high versatility.

[0011] Furthermore, in step a, a plane is selected as the parameterized model for each of the multiple sets of point cloud data formed by segmentation, and n sampling points are randomly selected from each set of data. If the n sampling points are suitable for the plane model and the deviation value of the n sampling points is less than a preset threshold, the point cloud data of the corresponding set is recorded. Finally, the best point cloud data is selected from the recorded multiple sets of point cloud data to serve as the derived plane equation of the ground. This setting facilitates the acquisition of the highest priority plane equation of the ground.

[0012] Furthermore, in step c, the direction of the rotation axis vector is... The planes they lie in are perpendicular, assuming Then the rotation axis vector: And according to the formula Calculate the angle θ between the axes of rotation. With this setup, the calculation of the included angle θ of the rotation axis is simple and the method has high computational efficiency.

[0013] Furthermore, the following steps are also included: d. Repeatedly collect point cloud data acquired by the point cloud camera N times, and perform the data processing from steps a to c. Then, average the calculated rotation axis and the rotation axis angle θ to obtain the final rotation axis and rotation axis angle θ data, which will be used as correction and compensation data for the point cloud camera in actual scene applications. This setting facilitates obtaining better temperature rotation neutralization angle calibration data, and in the subsequent calculations, converts the data collected by the area array radar into data in the world coordinate system, unifying the two data.

[0014] A storage medium stores a computer program configured to implement the AGV magnetic tracking offset calculation method when called by a processor. With this configuration, the storage medium of the present invention allows a point cloud camera to capture point cloud data of the environment, analyze the point cloud data to obtain the plane equation of the ground, and calculate the rotation axis and the included angle θ relative to the world coordinate system data. Finally, the calculated rotation axis and the included angle θ are used as position deviation compensation during on-site environmental photography. This effectively eliminates the need for multiple manual adjustments to the point cloud camera's position deviation, making debugging convenient. Furthermore, it is applicable to AGV vehicles equipped with different point cloud camera mounts, exhibiting high versatility. Attached Figure Description

[0015] Figure 1 A flowchart of a camera calibration method based on area array radar point cloud data. Detailed Implementation

[0016] The technical solution of the present invention is described below with reference to the accompanying drawings:

[0017] Example 1:

[0018] See Figure 1 The camera calibration method based on area array radar point cloud data of the present invention is characterized by comprising the following steps:

[0019] a. The point cloud data acquired by the point cloud camera is segmented according to spatial, ensemble, and texture features, so that the point clouds in the same segmented area have similar features as described above. Then, the segmented point cloud data is verified by plane equation calculation. For example, the segmented point cloud data is verified by using the Random Sample Consensus (RANSAC) algorithm as the core to obtain the plane equation of the ground (AX+BY+CZ+D=0).

[0020] b. Obtain the normal vector of the ground based on the plane equation of the ground (AX+BY+CZ+D=0). The coordinates are (A, B, C).

[0021] c. Use the ground's normal vector By performing a cross product with the Z-axis unit vector (0, 0, 1) of the world coordinate system, we can obtain the rotation axis between the two planes and the angle θ between the rotation axes.

[0022] Compared with existing technologies, the camera calibration method based on area array radar point cloud data of the present invention uses a point cloud camera to capture point cloud data of the environment, then uses the point cloud data to analyze and obtain the plane equation of the ground, and calculates the rotation axis and the included angle θ of the rotation axis relative to the world coordinate system data. Finally, the calculated rotation axis and the included angle θ are used as position deviation compensation when capturing images of the on-site environment. This effectively eliminates the need for manual adjustment of the point cloud camera position deviation, making debugging convenient. Moreover, it is applicable to AGV vehicles with different point cloud camera mountings, and has high versatility.

[0023] In one embodiment, in step a, a plane (SACMODEL_PLANE) is selected as the parameterized model for each of the multiple sets of point cloud data formed by segmentation, and n sampling points are randomly selected from each set of data. If the n sampling points are suitable for the plane model and the deviation value of the n sampling points is less than a preset threshold (the deviation value can be calculated by the mean, variance, etc.), then the point cloud data of the corresponding set is recorded. Finally, the best-choice point cloud data is selected from the recorded multiple sets of point cloud data to serve as the derived plane equation of the ground. This setting facilitates the acquisition of the highest-priority plane equation of the ground.

[0024] In one embodiment, in step c, the direction of the rotation axis vector is... The planes they lie in are perpendicular, assuming Then the rotation axis vector: And according to the formula Calculate the angle θ between the axes of rotation. With this setup, the calculation of the included angle θ of the rotation axis is simple and the method has high computational efficiency.

[0025] In one embodiment, the method further includes the following steps: d. Repeatedly collect point cloud data acquired by the point cloud camera N times, preferably N≥100, and perform the data processing from steps a to c. Then, average the calculated rotation axis and the included angle θ to obtain the final rotation axis and included angle θ data, which serves as correction and compensation data for the point cloud camera in actual application scenarios. By setting it up in this way, repeatedly collecting data N times and then averaging the calculation results improves the characteristics of data drift of the area array radar. After obtaining stable rotation axis and angle calibration data, the data acquired by the area array radar can be converted into data in the world coordinate system in the subsequent calculations, unifying the two sets of data.

[0026] Example 2:

[0027] See Figure 1The storage medium of the present invention stores a computer program configured to implement the AGV magnetic tracking offset calculation method of Embodiment 1 when called by a processor. With this configuration, the storage medium of the present invention uses a point cloud camera to capture point cloud data of the environment, then uses the point cloud data to analyze and obtain the plane equation of the ground, and calculates the rotation axis and the included angle θ of the rotation axis relative to the world coordinate system data. Finally, the calculated rotation axis and the included angle θ are used as position deviation compensation during on-site environmental shooting, thereby effectively eliminating the need for manual adjustment of the point cloud camera's position deviation, making debugging convenient, and applicable to AGV vehicles with different point cloud camera mountings, thus having high versatility.

[0028] Based on the disclosure and teachings of the foregoing specification, those skilled in the art can make changes and modifications to the above embodiments. Therefore, the present invention is not limited to the specific embodiments disclosed and described above, and some modifications and changes to the present invention should also fall within the protection scope of the claims of the present invention. Furthermore, although some specific terms are used in this specification, these terms are only for convenience of explanation and do not constitute any limitation on the present invention.

Claims

1. A camera calibration method based on point cloud data from a radar array, characterized in that, Includes the following steps: a. The point cloud data acquired by the point cloud camera is segmented according to the spatial, ensemble and texture features. Then, the segmented point cloud data is used to verify the plane equation to obtain the plane equation of the ground (AX+BY+CZ+D=0). b. Obtain the normal vector of the ground based on the plane equation of the ground (AX+BY+CZ+D=0). The values ​​are (A, B, C); c. Use the ground's normal vector By performing a cross product with the unit vector (0, 0, 1) of the Z-axis of the world coordinate system, we can obtain the rotation axis between the two planes and the angle θ between the rotation axes. In step c, the direction of the rotation axis vector is... The planes they lie in are perpendicular, assuming Then the rotation axis vector: And according to the formula Calculate the angle θ between the axes of rotation. The method also includes the following steps: d. Repeatedly collect point cloud data acquired by the point cloud camera N times, and perform the data processing from steps a to c. Then, average the calculated rotation axis and the rotation axis angle θ to obtain the final rotation axis and rotation axis angle θ data, which will be used as correction and compensation data for the point cloud camera in actual scene applications.

2. The camera calibration method based on area array radar point cloud data according to claim 1, characterized in that, In step a, a plane is selected as the parameterized model for each of the multiple sets of point cloud data formed by segmentation, and n sampling points are randomly selected from each set of data. If the n sampling points are suitable for the plane model and the deviation value of the n sampling points is less than a preset threshold, the point cloud data of the corresponding set is recorded. Finally, the optimal point cloud data is selected from the recorded multiple sets of point cloud data to serve as the derived plane equation of the ground.

3. A storage medium, characterized in that, The storage medium stores a computer program configured to implement the camera calibration method based on area array radar point cloud data as described in claim 1 or 2 when invoked by a processor.

Images