Three-dimensional geometric reconstruction camera calibration method and system

Abstract

The invention discloses a three-dimensional geometric reconstruction camera calibration method and system, and relates to the technical field of three-dimensional geometric reconstruction, and the method comprises the steps: S1, image collection; s2, automatic calibration point selection; s3, estimating camera parameters; s4, distortion correction is carried out; and S5, performing three-dimensional reconstruction verification. According to the three-dimensional geometric reconstruction camera calibration method and system, a calibration point does not need to be selected manually, the calibration process is simplified, the complexity and time cost of the calibration process are reduced, a high-precision calibration result can be obtained without shooting a plurality of plane images and carrying out a plurality of optimization iterations, and the calibration efficiency is improved. The calculation amount and the time cost are further reduced, and camera calibration can be quickly and accurately realized; in addition, nonlinear distortion of a camera lens is fully considered, the calibration precision is further improved, and the method is particularly suitable for the situation that a wide-angle lens is used or an image contains large distortion.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional geometric reconstruction, and in particular to a three-dimensional geometric reconstruction camera calibration method and system. Background Art

[0002] 3D geometric reconstruction refers to the process of restoring the 3D geometric structure of an object or scene from 2D image information. This technology is applied in computer vision, medical image processing, virtual reality, augmented reality and other fields. Specifically, 3D geometric reconstruction can restore the 3D coordinates of each pixel relative to the camera through photos from multiple angles, thereby restoring the 3D shape of the object or scene.

[0003] Camera calibration is a key step in 3D geometric reconstruction. It refers to determining the internal and external parameters of the camera in order to map points in a 2D image to points in 3D space. Currently, common camera calibration methods include: Zhang Zhengyou calibration method, Bouguet improved method, calibration method using optimization algorithm, and calibration method using camera transformation matrix.

[0004] However, in actual use, the existing 3D geometric reconstruction camera calibration methods usually require manual selection of calibration points, which makes the calibration process more cumbersome, increases the complexity and time cost of the calibration process, and in order to obtain higher-precision calibration results, it is often necessary to take more plane images and perform more optimization iterations, which further increases the amount of calculation and time cost; in addition, since the existing methods generally do not consider the nonlinear distortion of the camera lens, such as radial distortion and tangential distortion, this affects the calibration accuracy, especially when using a wide-angle lens or the image contains large distortion.

[0005] Therefore, there is an urgent need to improve this shortcoming. The present invention studies and improves the existing technology and its shortcomings, and provides a 3D geometric reconstruction camera calibration method and system. Summary of the invention

[0006] The object of the present invention is to provide a 3D geometric reconstruction camera calibration method and system to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] On the one hand, a 3D geometric reconstruction camera calibration method is provided, comprising the following steps:

[0009] S1. Image acquisition: Use a camera to capture a set of images of calibration objects (such as stereo calibration objects) containing known three-dimensional information, and the images contain calibration objects at different angles and viewing angles, so as to perform subsequent accurate three-dimensional reconstruction;

[0010] S2, automatic calibration point selection: Use image processing algorithms to automatically detect and select feature points (such as corners, edges, etc.) on the calibration object as calibration points. These calibration points will be used for subsequent 3D reconstruction and camera calibration;

[0011] S3, camera parameter estimation: according to the selected calibration points, the camera's intrinsic parameters (such as focal length, optical center position, distortion coefficient, etc.) and extrinsic parameters (such as the position and posture of the camera relative to the calibration object) are estimated using a mathematical optimization algorithm;

[0012] S4, distortion correction: according to the estimated distortion coefficient, the image is corrected pixel by pixel to eliminate radial distortion caused by the camera lens (caused by the change in curvature of the lens along the radial direction, usually manifested as barrel distortion or pincushion distortion, wherein barrel distortion is caused by the light rays far from the optical axis bending more inward, and pincushion distortion is caused by the light rays far from the optical axis bending more outward) and tangential distortion (caused by the non-parallelism of the lens and the imaging plane, resulting in displacement of the pixel position);

[0013] S5. 3D reconstruction verification: Use the calibrated camera parameters and calibration point information to perform 3D reconstruction of the calibrated object and verify the precision and accuracy of the reconstruction results.

[0014] Furthermore, in step S1, the specific process of camera shooting is as follows:

[0015] Clarify the purpose and scene of the shooting: select the camera parameter settings according to the subject of the shooting (calibration object), and analyze the shooting environment, considering the lighting conditions of the shooting environment (such as natural light, artificial light), the complexity of the background, and whether mobile shooting is required;

[0016] Adjust the focal length and viewing angle: Choose a wide-angle lens with a shorter focal length to capture wide scenes, or choose a telephoto lens with a longer focal length to capture scenes from a distance or where the subject needs to be highlighted. Adjust the angle and height of the camera according to your shooting needs to obtain the best visual effect.

[0017] Set the aperture and shutter speed: In bright light, choose a smaller aperture (larger aperture value) to achieve a greater depth of field, so that both the foreground and background in the picture remain clear, and use a faster shutter speed to freeze the moment in a dynamic scene. In dim light, increase the aperture (smaller aperture value) to increase the shutter speed to prevent image blur, and reduce the shutter speed to increase exposure, but be careful to prevent the image from being blurred due to camera shake;

[0018] Adjust ISO sensitivity: In bright light, choose a low ISO value to reduce noise and improve image quality. In dim light, increase the ISO value to increase the camera's sensitivity.

[0019] White balance setting: Adjust the camera's white balance setting according to the light conditions of the shooting environment to obtain a more realistic color reproduction effect;

[0020] Focus mode selection: Select the appropriate focus mode according to your shooting needs, such as single autofocus, continuous autofocus, etc., to ensure that the subject is always clear;

[0021] Actual shooting: Perform actual shooting under the preset camera parameters to observe whether the image effect meets the requirements, and fine-tune and optimize the camera parameters based on the actual shooting effect.

[0022] Furthermore, in step S2, calibrating feature points on the object includes:

[0023] Corner feature: a point in the image where the brightness changes dramatically, with rotation invariance and scale invariance;

[0024] Edge feature: The boundary line between different areas in the image has characteristics such as maximum gradient or gray value mutation.

[0025] Furthermore, in step S2, the image processing algorithm includes a corner detection algorithm, an edge detection algorithm and a feature matching algorithm;

[0026] The corner point detection algorithm adopts Shi-Tomasi corner point detection, determines the corner point by calculating the value of the corner point response function, and introduces an adaptive threshold to select the optimal corner point;

[0027] The edge detection algorithm adopts Canny edge detection, performs noise reduction through Gaussian smoothing filter, then uses a pair of convolution arrays to calculate edge gradient and direction, then uses non-maximum suppression to remove non-edge lines, and finally uses hysteresis threshold to detect and connect edges.

[0028] Furthermore, the feature matching algorithm is used to identify points of the same name between two or more images, specifically including:

[0029] Template matching: A pixel-based matching method used to search for the location of a template image in a large image. It only performs parallel translation matching and is invalid for matching targets that are rotated or resized. The specific operation is to slide the template image on the input image and compare the template image with the corresponding sub-region of the input image at each position.

[0030] SIFT feature matching: an algorithm for detecting local features, which is scale-invariant and rotation-invariant. Even if the image brightness or shooting angle is changed, good detection results can still be obtained. Specific operation: by finding the feature points in an image and their related scale and direction descriptions, the features are obtained and image feature point matching is performed;

[0031] FLANN feature matching: a fast nearest neighbor search algorithm suitable for matching tasks of large-scale image data. Specific operations: perform nearest neighbor search in large data sets and high-dimensional features.

[0032] Furthermore, in step S3, the camera intrinsic parameters and extrinsic parameters are as follows:

[0033] Intrinsic parameters: describe the internal geometric characteristics of the camera, including focal length (usually in pixels), optical center position (center point of the image), and distortion coefficients (including radial distortion and tangential distortion, etc.);

[0034] External parameters: describe the position and attitude of the camera relative to the world coordinate system, including the rotation matrix and translation vector. The rotation matrix describes the rotation angle of the camera, and the translation vector describes the translation distance of the camera.

[0035] Furthermore, in step S3, the mathematical optimization algorithm estimates the intrinsic and extrinsic parameters of the camera by minimizing the error between the calibration point and the reconstructed three-dimensional point, and the mathematical optimization algorithm includes a least squares method and an iterative closest point algorithm (ICP);

[0036] Least squares method: Find the best function match for the data by minimizing the sum of squares of the error. In camera calibration, the least squares method is used to solve a system of linear equations to obtain the camera's internal and external parameters and quickly provide a preliminary estimate of the camera's parameters. Specific operations: When multiple matching 3D points and their 2D projection points on the image are collected, a series of linear equations can be constructed, and then the least squares method is used to solve the least squares solution of this linear system to obtain the camera's internal and external parameter matrices.

[0037] Iterative Closest Point (ICP) algorithm: a point set to point set registration method, which is based on the least squares method to find the optimal match. In camera calibration, it is used to iteratively adjust the external parameters of the camera to minimize the error between the reconstructed 3D points and the calibration points. The specific operation is: first calculate the correspondence between two point sets (one from the calibration object and the other from the reconstructed 3D points), and then calculate the optimal rigid body transformation (rotation and translation) to minimize the error between the two point sets. This process is iterative until a certain convergence criterion is reached.

[0038] Furthermore, in step S4, the specific operation steps of distortion correction are as follows:

[0039] Obtain distortion coefficients: Based on the estimated distortion coefficients, obtain radial distortion coefficients (such as k1, k2, k3, etc.) and tangential distortion coefficients (such as p1, p2, etc.);

[0040] Constructing a distortion correction model: Based on the distortion coefficients and camera internal parameters, a mathematical model for distortion correction is constructed. The model describes how to map from the pixel position in the distorted image to the pixel position in the corrected image.

[0041] Apply the distortion correction model: For each pixel in the input image, use the distortion correction model to calculate its position in the corrected image, and if the corrected position is a floating point number, use an interpolation algorithm (such as bilinear interpolation) to calculate the pixel value at that position;

[0042] Generate the corrected image: Traverse all pixels in the input image and repeat the above steps to generate the corrected image.

[0043] On the other hand, a 3D geometric reconstruction camera calibration system is provided, which is applied to the 3D geometric reconstruction camera calibration method as described above and applied to a server, including:

[0044] An image acquisition module, used to acquire images of the calibration object containing known three-dimensional information;

[0045] Automatic selection module, used to automatically detect and select feature points on the calibration object as calibration points;

[0046] Parameter estimation module, used to estimate the intrinsic and extrinsic parameters of the camera;

[0047] A distortion correction module, used for performing pixel-by-pixel correction processing on the image according to the estimated distortion coefficient;

[0048] A 3D reconstruction module is used to perform 3D reconstruction of the calibrated object using the calibrated camera parameters and calibration point information;

[0049] The verification module is used to verify the precision and accuracy of the reconstruction results.

[0050] Furthermore, the processor is used to execute computer instructions to enable the system to execute: an image acquisition module, an automatic selection module, a parameter estimation module, a distortion correction module, a three-dimensional reconstruction module and a verification module;

[0051] The memory is used to store computer instructions. A computer program is stored in the memory. When the computer program is executed by the processor, the three-dimensional geometric reconstruction camera calibration method as described above is implemented.

[0052] The present invention provides a 3D geometric reconstruction camera calibration method and system, which has the following beneficial effects:

[0053] The present invention does not need to manually select calibration points, simplifies the calibration process, thereby reducing the complexity and time cost of the calibration process, and can obtain higher-precision calibration results without taking a large number of plane images and performing a large number of optimization iterations, further reducing the amount of calculation and time cost, and helping to quickly and accurately achieve camera calibration; in addition, the nonlinear distortion of the camera lens is fully taken into account, and the calibration accuracy is further improved, which is particularly suitable for situations where a wide-angle lens is used or the image contains large distortion. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 A schematic flow chart of the steps of a camera calibration method for 3D geometric reconstruction according to the present invention;

[0055] Figure 2 The present invention is a logic block diagram of a three-dimensional geometric reconstruction camera calibration system. DETAILED DESCRIPTION

[0056] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0057] like Figure 1 As shown, a 3D geometric reconstruction camera calibration method includes the following steps:

[0058] S1. Image acquisition: Use a camera to capture a set of images of calibration objects (such as stereo calibration objects) containing known three-dimensional information, and the images contain calibration objects at different angles and viewing angles, so as to perform subsequent accurate three-dimensional reconstruction;

[0059] In this embodiment, the specific process of camera shooting is as follows:

[0060] Clarify the purpose and scene of the shooting: select the camera parameter settings according to the subject of the shooting (calibration object), and analyze the shooting environment, considering the lighting conditions of the shooting environment (such as natural light, artificial light), the complexity of the background, and whether mobile shooting is required;

[0061] Adjust the focal length and viewing angle: Choose a wide-angle lens with a shorter focal length to capture wide scenes, or choose a telephoto lens with a longer focal length to capture scenes from a distance or where the subject needs to be highlighted. Adjust the angle and height of the camera according to your shooting needs to obtain the best visual effect.

[0062] Set the aperture and shutter speed: In bright light, choose a smaller aperture (larger aperture value) to achieve a greater depth of field, so that both the foreground and background in the picture remain clear, and use a faster shutter speed to freeze the moment in a dynamic scene. In dim light, increase the aperture (smaller aperture value) to increase the shutter speed to prevent image blur, and reduce the shutter speed to increase exposure, but be careful to prevent the image from being blurred due to camera shake;

[0063] Adjust ISO sensitivity: In bright light, choose a low ISO value to reduce noise and improve image quality. In dim light, increase the ISO value to increase the camera's sensitivity.

[0064] White balance setting: Adjust the camera's white balance setting according to the light conditions of the shooting environment to obtain a more realistic color reproduction effect;

[0065] Focus mode selection: Select the appropriate focus mode according to your shooting needs, such as single autofocus, continuous autofocus, etc., to ensure that the subject is always clear;

[0066] Actual shooting: Perform actual shooting under the preset camera parameters to observe whether the image effect meets the requirements, and fine-tune and optimize the camera parameters according to the actual shooting effect;

[0067] S2. Automatic calibration point selection: Use image processing algorithms to automatically detect and select feature points on the calibration object as calibration points. These calibration points will be used for subsequent 3D reconstruction and camera calibration. The feature points on the calibration object include:

[0068] Corner feature: a point in the image where the brightness changes dramatically, with rotation invariance and scale invariance;

[0069] Edge features: The boundary lines between different areas in an image have characteristics such as maximum gradient or gray value mutation;

[0070] In this embodiment, the image processing algorithm includes a corner detection algorithm, an edge detection algorithm and a feature matching algorithm;

[0071] The corner point detection algorithm adopts Shi-Tomasi corner point detection, determines the corner point by calculating the value of the corner point response function, and introduces an adaptive threshold to select the optimal corner point;

[0072] The edge detection algorithm uses Canny edge detection, performs noise reduction through a Gaussian smoothing filter, then uses a pair of convolution arrays to calculate edge gradients and directions, then uses non-maximum suppression to remove non-edge lines, and finally uses a hysteresis threshold to detect and connect edges;

[0073] The feature matching algorithm is used to identify points of the same name between two or more images, specifically including:

[0074] Template matching: A pixel-based matching method used to search for the location of a template image in a large image. It only performs parallel translation matching and is invalid for matching targets that are rotated or resized. The specific operation is to slide the template image on the input image and compare the template image with the corresponding sub-region of the input image at each position.

[0075] SIFT feature matching: an algorithm for detecting local features, which is scale-invariant and rotation-invariant. Even if the image brightness or shooting angle is changed, good detection results can still be obtained. Specific operation: by finding the feature points in an image and their related scale and direction descriptions, the features are obtained and image feature point matching is performed;

[0076] FLANN feature matching: a fast nearest neighbor search algorithm suitable for matching tasks of large-scale image data. Specific operations: nearest neighbor search in large data sets and high-dimensional features;

[0077] S3, Camera parameter estimation: Based on the selected calibration points, the camera's intrinsic parameters (such as focal length, optical center position, distortion coefficient, etc.) and extrinsic parameters (such as the position and posture of the camera relative to the calibration object) are estimated using a mathematical optimization algorithm; the camera's intrinsic and extrinsic parameters are as follows:

[0078] Intrinsic parameters: describe the internal geometric characteristics of the camera, including focal length (usually in pixels), optical center position (center point of the image), and distortion coefficients (including radial distortion and tangential distortion, etc.);

[0079] External parameters: describe the position and posture of the camera relative to the world coordinate system, including the rotation matrix and translation vector. The rotation matrix describes the rotation angle of the camera, and the translation vector describes the translation distance of the camera.

[0080] In this embodiment, the mathematical optimization algorithm estimates the intrinsic parameters and extrinsic parameters of the camera by minimizing the error between the calibration point and the reconstructed three-dimensional point, and the mathematical optimization algorithm includes the least squares method and the iterative closest point algorithm (ICP);

[0081] Least squares method: Find the best function match for the data by minimizing the sum of squares of the error. In camera calibration, the least squares method is used to solve a system of linear equations to obtain the camera's internal and external parameters and quickly provide a preliminary estimate of the camera's parameters. Specific operations: When multiple matching 3D points and their 2D projection points on the image are collected, a series of linear equations can be constructed, and then the least squares method is used to solve the least squares solution of this linear system to obtain the camera's internal and external parameter matrices.

[0082] Iterative Closest Point (ICP): A point-to-point registration method that finds the best match based on the least squares method. In camera calibration, it is used to iteratively adjust the camera's extrinsic parameters to minimize the error between the reconstructed 3D points and the calibration points. The specific operation is: first calculate the correspondence between two point sets (one from the calibration object and the other from the reconstructed 3D points), then calculate the optimal rigid body transformation (rotation and translation) to minimize the error between the two point sets, and this process is iterative until a certain convergence criterion is reached;

[0083] S4, distortion correction: according to the estimated distortion coefficient, the image is corrected pixel by pixel to eliminate radial distortion caused by the camera lens (caused by the change in curvature of the lens along the radial direction, usually manifested as barrel distortion or pincushion distortion, wherein barrel distortion is caused by the light rays far from the optical axis bending more inward, and pincushion distortion is caused by the light rays far from the optical axis bending more outward) and tangential distortion (caused by the non-parallelism of the lens and the imaging plane, resulting in displacement of the pixel position); the specific operation steps of distortion correction are as follows:

[0084] Obtain distortion coefficients: Based on the estimated distortion coefficients, obtain radial distortion coefficients (such as k1, k2, k3, etc.) and tangential distortion coefficients (such as p1, p2, etc.);

[0085] Constructing a distortion correction model: Based on the distortion coefficients and camera internal parameters, a mathematical model for distortion correction is constructed. The model describes how to map from the pixel position in the distorted image to the pixel position in the corrected image.

[0086] Apply the distortion correction model: For each pixel in the input image, use the distortion correction model to calculate its position in the corrected image, and if the corrected position is a floating point number, use an interpolation algorithm (such as bilinear interpolation) to calculate the pixel value at that position;

[0087] Generate the corrected image: traverse all pixels in the input image and repeat the above steps to generate the corrected image;

[0088] S5, 3D reconstruction verification: Use the calibrated camera parameters and calibration point information to perform 3D reconstruction of the calibration object and verify the precision and accuracy of the reconstruction result;

[0089] In this embodiment, evaluation indicators such as reprojection error and average reprojection error are used to evaluate the accuracy of the reconstruction result. At the same time, the accuracy of the reconstruction result is verified by comparing it with other high-precision measurement methods (such as laser scanners, etc.).

[0090] like Figure 2 As shown, a 3D geometric reconstruction camera calibration system is applied to the 3D geometric reconstruction camera calibration method as described above, and is applied to a server, comprising:

[0091] An image acquisition module, used to acquire images of the calibration object containing known three-dimensional information;

[0092] Automatic selection module, used to automatically detect and select feature points on the calibration object as calibration points;

[0093] Parameter estimation module, used to estimate the intrinsic and extrinsic parameters of the camera;

[0094] A distortion correction module, used for performing pixel-by-pixel correction processing on the image according to the estimated distortion coefficient;

[0095] A 3D reconstruction module is used to perform 3D reconstruction of the calibrated object using the calibrated camera parameters and calibration point information;

[0096] A verification module, used to verify the precision and accuracy of the reconstruction results;

[0097] A processor, used to execute computer instructions to enable the system to execute: an image acquisition module, an automatic selection module, a parameter estimation module, a distortion correction module, a three-dimensional reconstruction module and a verification module;

[0098] The memory is used to store computer instructions. A computer program is stored in the memory. When the computer program is executed by the processor, the three-dimensional geometric reconstruction camera calibration method as described above is implemented.

[0099] The embodiments of the present invention are given for the purpose of illustration and description, and are not intended to be exhaustive or to limit the invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

Claims

1. A camera calibration method for 3D geometric reconstruction, characterized in that: The following steps are involved: S1. Image acquisition: Use a camera to capture a set of images of calibration objects containing known three-dimensional information, and the images contain calibration objects at different angles and viewing angles; S2, automatic calibration point selection: using image processing algorithms to automatically detect and select feature points on the calibration object as calibration points; S3, camera parameter estimation: according to the selected calibration points, the camera's intrinsic and extrinsic parameters are estimated using a mathematical optimization algorithm; S4, distortion correction: according to the estimated distortion coefficient, the image is corrected pixel by pixel to eliminate the radial distortion and tangential distortion caused by the camera lens; S5. 3D reconstruction verification: Use the calibrated camera parameters and calibration point information to perform 3D reconstruction of the calibrated object and verify the precision and accuracy of the reconstruction result.

2. A 3D geometric reconstruction camera calibration method according to claim 1, characterized in that: In step S1, the specific process of camera shooting is as follows: Clarify the purpose and scene of the shooting: select the camera parameter settings according to the subject of the shooting, and analyze the shooting environment, considering the lighting conditions of the shooting environment, the complexity of the background, and whether mobile shooting is required; Adjust the focal length and viewing angle: Choose a wide-angle lens with a shorter focal length to capture wide scenes, or choose a telephoto lens with a longer focal length to capture scenes from a distance or where the subject needs to be highlighted. Adjust the angle and height of the camera according to your shooting needs to obtain the best visual effect. Set the aperture and shutter speed: In bright light, choose a smaller aperture to get a greater depth of field, and use a faster shutter speed to freeze the moment in a dynamic scene, while in dim light, increase the aperture to increase the shutter speed, and decrease the shutter speed to increase the exposure; Adjust ISO sensitivity: In bright light, choose a low ISO value, and in dim light, increase the ISO value; White balance setting: adjust the camera's white balance setting according to the light conditions of the shooting environment; Focus mode selection: Select the appropriate focus mode according to your shooting needs to ensure that the subject is always clear; Actual shooting: Perform actual shooting under the preset camera parameters, observe whether the image effect meets the requirements, and fine-tune and optimize the camera parameters based on the actual shooting effect.

3. The 3D geometric reconstruction camera calibration method according to claim 1, characterized in that: In the step S2, calibrating the feature points on the object includes: Corner feature: a point in the image where the brightness changes dramatically, with rotation invariance and scale invariance; Edge feature: The boundary line between different areas in the image has the characteristics of gradient maximum or gray value mutation.

4. A 3D geometric reconstruction camera calibration method according to claim 3, characterized in that: In step S2, the image processing algorithm includes a corner detection algorithm, an edge detection algorithm and a feature matching algorithm; The corner point detection algorithm adopts Shi-Tomasi corner point detection, determines the corner point by calculating the value of the corner point response function, and introduces an adaptive threshold to select the optimal corner point; The edge detection algorithm adopts Canny edge detection, performs noise reduction through Gaussian smoothing filter, then uses a pair of convolution arrays to calculate edge gradient and direction, then uses non-maximum suppression to remove non-edge lines, and finally uses hysteresis threshold to detect and connect edges.

5. A 3D geometric reconstruction camera calibration method according to claim 4, characterized in that: The feature matching algorithm is used to identify points of the same name between two or more images, specifically including: Template matching: Use a template image to slide over the input image and compare the template image with its corresponding sub-region of the input image at each location; SIFT feature matching: by finding the feature points in an image and their descriptions of scale and direction, the features are obtained and image feature points are matched; FLANN feature matching: Nearest neighbor search in large datasets and high-dimensional features.

6. The 3D geometric reconstruction camera calibration method according to claim 1, characterized in that: In step S3, the camera intrinsic parameters and extrinsic parameters are as follows: Intrinsic parameters: describe the internal geometric characteristics of the camera, including focal length, optical center position, and distortion coefficient; External parameters: describe the position and attitude of the camera relative to the world coordinate system, including the rotation matrix and translation vector. The rotation matrix describes the rotation angle of the camera, and the translation vector describes the translation distance of the camera.

7. The 3D geometric reconstruction camera calibration method according to claim 1, characterized in that: In step S3, the mathematical optimization algorithm estimates the intrinsic parameters and extrinsic parameters of the camera by minimizing the error between the calibration point and the reconstructed three-dimensional point, and the mathematical optimization algorithm includes a least squares method and an iterative closest point algorithm; Least squares method: used to solve the linear equations, obtain the intrinsic and extrinsic parameters of the camera, and quickly provide preliminary camera parameter estimates; Iterative closest point algorithm: used to iteratively adjust the external parameters of the camera to minimize the error between the reconstructed 3D points and the calibration points.

8. The 3D geometric reconstruction camera calibration method according to claim 1, characterized in that: In step S4, the specific operation steps of distortion correction are as follows: Obtaining distortion coefficients: based on the estimated distortion coefficients, obtaining radial distortion coefficients and tangential distortion coefficients; Constructing a distortion correction model: Constructing a mathematical model for distortion correction based on the distortion coefficient and camera internal parameters; Apply the distortion correction model: For each pixel in the input image, use the distortion correction model to calculate its position in the corrected image, and use the interpolation algorithm to calculate the pixel value at that position if the corrected position is a floating point number; Generate the corrected image: Traverse all pixels in the input image and repeat the above steps to generate the corrected image.

9. A 3D geometric reconstruction camera calibration system, applied to the 3D geometric reconstruction camera calibration method according to any one of claims 1 to 8, characterized in that: Applicable to servers, including: An image acquisition module, used to acquire images of the calibration object containing known three-dimensional information; Automatic selection module, used to automatically detect and select feature points on the calibration object as calibration points; Parameter estimation module, used to estimate the intrinsic and extrinsic parameters of the camera; A distortion correction module, used for performing pixel-by-pixel correction processing on the image according to the estimated distortion coefficient; A 3D reconstruction module is used to perform 3D reconstruction of the calibrated object using the calibrated camera parameters and calibration point information; The verification module is used to verify the precision and accuracy of the reconstruction results.

10. The 3D geometric reconstruction camera calibration system according to claim 9, characterized in that: include: A processor, used to execute computer instructions to enable the system to execute: an image acquisition module, an automatic selection module, a parameter estimation module, a distortion correction module, a three-dimensional reconstruction module and a verification module; A memory for storing computer instructions, wherein a computer program is stored in the memory, and when the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images