Single-camera rapid space three-dimensional reconstruction measurement method
Through the single-camera rapid spatial three-dimensional reconstruction measurement method, the imaging distortion parameters and feature extraction technology are used to solve the problems of high cost of multi-eye vision methods and difficult to maintain relative spatial relationships, and high-precision three-dimensional spatial reconstruction is achieved.
Patent Information
- Application Number
- CN202411308755.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-19
- Publication Date
- 2024-11-08
AI Technical Summary
In existing automatic navigation and spatial mapping technologies, multi-eye vision methods are costly and the relative spatial relationship between cameras is difficult to maintain for a long time, resulting in accuracy and reliability problems in three-dimensional spatial measurements.
Using a single-camera fast space three-dimensional reconstruction measurement method, the imaging distortion parameters of the camera probe are obtained, the characteristic key points and edge information in the image are extracted, and the three-dimensional coordinate reconstruction is carried out, and the camera calibration is used to compensate for lens distortion. The measurement coordinate system is fixed to ensure data consistency.
It reduces the cost of use, improves the accuracy and reliability of three-dimensional space reconstruction, can more accurately understand the geometry and position of the target object, and enhances the accuracy and reliability of the reconstruction results.
Smart Images

Figure CN118918270A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of automatic navigation and space mapping, and in particular to a single-camera fast space three-dimensional reconstruction measurement method. Background Art
[0002] Automatic navigation and spatial mapping technology is an important technology for assisting or independent decision-making in the fields of vehicles and aircraft. At present, automatic navigation and spatial mapping technology mainly uses laser radar or visual image methods to measure and analyze three-dimensional space. Since laser radar is only point cloud data, it lacks characteristic information such as color, and the measurement distance is affected by laser power and laser scanning speed, so it is difficult to perform large-scale high-speed measurement. In addition, since there will be corresponding moving platform interference during the movement of vehicles and aircraft, the data collected by vehicles and aircraft are difficult to reconstruct in three-dimensional space during the movement. Visual images can obtain image information of the three-dimensional environment by using camera images, which is used to analyze information such as interference objects and signal indications in three-dimensional space. It is the focus of the development of automatic navigation technology in the future. However, in terms of three-dimensional space measurement, it is still mainly based on the principle of multi-eye vision, using two or more camera probes for three-dimensional space measurement. Although multi-eye vision has the advantage of high precision, it is costly and the relative spatial relationship between cameras is difficult to maintain for a long time during operation. Summary of the invention
[0003] In view of the above-mentioned deficiencies in the prior art, the present invention provides a single-camera fast spatial three-dimensional reconstruction measurement method, which solves the problem of high cost of using multiple camera probes in the prior art.
[0004] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: a single-camera fast spatial three-dimensional reconstruction measurement method, comprising the following steps:
[0005] S1, obtaining imaging distortion parameters of the camera probe;
[0006] Wherein, the camera probe is a single-camera probe, and the camera probe is installed on a motion platform;
[0007] S2, setting a measurement coordinate system, and acquiring continuous images captured by the camera probe in the measurement coordinate system;
[0008] S3, extracting features from continuous images to obtain feature key points and edge information;
[0009] S4, reconstructing the target object in the continuous image according to the feature key points and the imaging distortion parameters to obtain the three-dimensional coordinates of the feature key points, and segmenting the target object according to the edge information to obtain the edge segmentation result of the target object;
[0010] S5. Perform three-dimensional space reconstruction according to the three-dimensional coordinates of the feature key points and the edge segmentation results to obtain a three-dimensional space reconstruction result.
[0011] The beneficial effects of the above scheme are as follows: the present invention reduces the cost of use by applying a single-camera probe, and by extracting feature key points and edge information in the image, three-dimensional space reconstruction can be performed more accurately, which helps to improve the understanding of the geometric shape and position of the target object. The acquisition and application of imaging distortion parameters can compensate for image distortion caused by lens distortion of the camera probe, thereby improving the accuracy and reliability of the reconstruction results.
[0012] Furthermore, step S1 specifically includes:
[0013] Use the calibration plate and the cross to calibrate the camera probe at different distances to obtain the imaging distortion parameters;
[0014] The method of using a calibration plate and a cross to calibrate the camera probe at different distances is the Zhang Zhengyou calibration method, or the industrial close-range photogrammetry calibration method.
[0015] The beneficial effects of the above further solution are: the calibration plate and the cross can help accurately measure the distortion parameters of the camera, and can effectively compensate for the image distortion caused by the camera lens, thereby improving the accuracy and clarity of the image. By accurately calibrating the distortion parameters of the camera, the three-dimensional model can be reconstructed more accurately. After the distortion compensation, the feature points and edge information in the image will be more accurate, which helps to improve the accuracy of the three-dimensional reconstruction result.
[0016] Furthermore, in step S2, setting the measurement coordinate system specifically includes:
[0017] S21. Set at least 5 marking points in the measurement space;
[0018] S22, using a camera probe to capture images of the landmark at three different motion positions;
[0019] S23, using the resection method to solve the external parameter difference of the camera probe according to the image of the marker point;
[0020] S24, obtaining a corresponding relationship between a motion coordinate system of the camera probe and an imaging coordinate system of the camera probe according to the difference in external parameters;
[0021] The formula used to obtain the correspondence between the motion coordinate system of the camera probe and the imaging coordinate system of the camera probe is:
[0022]
[0023] Among them, P f represents the spatial coordinates of the camera probe at the shooting time, Pc The image of the camera probe is based on the coordinates of the optical center of the intersection solution of the marker point coordinates, R fc and T fc Represents the correspondence between the motion coordinate system of the camera probe and the imaging coordinate system of the camera probe;
[0024] S25. Setting a measurement coordinate system according to the correspondence between the motion coordinate system of the camera probe and the imaging coordinate system of the camera probe.
[0025] The beneficial effect of the above further solution is that the measurement coordinate system provides a unified reference frame, so that different measurement points, images or data can be compared and analyzed in a common coordinate system, which helps to ensure the accuracy and consistency of the measurement data.
[0026] Furthermore, in step S3, feature extraction is performed on the continuous images, and the methods used include edge extraction algorithm, SIFT algorithm and SURF algorithm.
[0027] Furthermore, in step S4, the target object in the continuous image is reconstructed according to the feature key points and the imaging distortion parameters, and the target object reconstruction formula used is:
[0028] A=[P i ' -1 -C i-1 C i -P i ′]
[0029] B=[P i ′-P i ' -1 ]
[0030] A*P i 3D =B
[0031] Among them, A and B represent parameter matrices, i represents the image sequence index number, P i 3D represents the three-dimensional coordinates of the feature key points in the i-th image, P i ′ represents the three-dimensional coordinates of the key feature points in the i-th image on the imaging plane after removing the imaging distortion, P i ' -1 represents the three-dimensional coordinates of the key points in the i-1th image on the imaging plane after removing the imaging distortion, C i represents the optical center coordinates of the camera in the i-th image, C i-1 Represents the coordinates of the optical center of the camera in the i-1th image.
[0032] Furthermore, the method for solving the three-dimensional coordinates of the feature key points on the imaging plane after removing the imaging distortion is the Zhang Zhengyou distortion model;
[0033] Zhang Zhengyou's distortion model is:
[0034]
[0035] in, and represents the two-dimensional coordinates of the imaging plane after the imaging distortion is removed from the feature key points, x and y represent the actual two-dimensional coordinates of the feature key points extracted from the acquired image, r represents the distance from the imaging point of the feature key point in the two-dimensional image to the imaging center, u0 and v0 represent the coordinates of the imaging optical center of the camera probe obtained by calibration in the imaging plane, k1 and k2 represent the imaging distortion parameters obtained using Zhang Zhengyou's calibration method.
[0036] Furthermore, the method for solving the three-dimensional coordinates of the feature key points on the imaging plane after removing the imaging distortion is an industrial photogrammetry model;
[0037] The industrial photogrammetry model is:
[0038]
[0039] in, and represents the two-dimensional coordinates of the feature key point in the imaging plane after the imaging distortion is removed, x and y represent the actual two-dimensional coordinates of the feature key point extracted from the acquired image, r represents the distance between the imaging point of the feature key point in the two-dimensional image and the imaging center, u0 and v0 represent the coordinates of the imaging optical center of the camera probe obtained by calibration in the imaging plane, A1, A2, A3, B1, B2, C1 and C2 represent the imaging distortion parameters obtained using the industrial photogrammetry calibration method.
[0040] Furthermore, in step S4, the target object in the continuous image is reconstructed according to the feature key points and the imaging distortion parameters to obtain the three-dimensional coordinates of the feature key points, which specifically includes:
[0041] S41, obtaining the coordinates of the two-dimensional points on the imaging plane after removing the imaging distortion according to the characteristic key points and the imaging distortion parameters;
[0042] S42, according to a conversion formula, converting the coordinates of the two-dimensional point on the imaging plane after removing the imaging distortion into the three-dimensional coordinates of the feature key point on the imaging plane after removing the imaging distortion;
[0043] The conversion formula is:
[0044]
[0045] Where P′ represents the three-dimensional coordinates of the key points on the imaging plane after removing the imaging distortion. and represents the two-dimensional coordinates of the imaging plane after removing the imaging distortion of the feature key points, K pz is the pixel size corresponding to the pixel, that is, the coefficient corresponding to the physical size of the image pixel on the optical sensor, R cam The relationship between the optical center, optical axis and imaging optical sensor in the imaging optical path, in the imaging system of ordinary industrial cameras The positive and negative signs in the ±f term are determined according to the definition of the optical center position, and f represents the focal length, that is, the actual physical size focal length of the imaging system;
[0046] S43, obtaining the three-dimensional coordinates of the feature key points according to the three-dimensional coordinates of the feature key points on the imaging plane after removing the imaging distortion.
[0047] The beneficial effects of the above further scheme are: Zhang Zhengyou distortion model can efficiently and accurately eliminate radial distortion and improve the quality of image imaging; the industrial photogrammetry model improves the accuracy of industrial production, speeds up production efficiency and reduces production costs.
[0048] Furthermore, in step S4, the method used for segmenting the target object is edge extraction based on digital image processing, or feature extraction, segmentation and recognition based on artificial intelligence.
[0049] Furthermore, step S5 specifically includes:
[0050] Determine whether the three-dimensional coordinates of the key feature points meet the point static evaluation conditions;
[0051] When the three-dimensional coordinates of the feature key points meet the point static evaluation conditions, three-dimensional space reconstruction is performed according to the three-dimensional coordinates of the feature key points and the edge segmentation results to obtain the three-dimensional space reconstruction results;
[0052] When the three-dimensional coordinates of the feature key points do not meet the point static evaluation conditions, reselect the feature key points;
[0053] The static evaluation conditions of the point are:
[0054]
[0055] Among them, P i 3D Represents the three-dimensional coordinates of the feature key points in the i-th image, Represents the three-dimensional coordinates of the key feature points in the i-1th image, Threshoud Static Represents the spatial feature stability threshold.
[0056] The above further solution has the beneficial effect that high-precision three-dimensional coordinates can more accurately reflect the shape and position of an object, thereby meeting the application requirements with high precision. Under static evaluation conditions, high-precision three-dimensional coordinates can provide more reliable data support and help make more accurate decisions. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] Figure 1 The figure is a flowchart of a single-camera fast spatial three-dimensional reconstruction measurement method. DETAILED DESCRIPTION
[0058] Now, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. It should be understood that the embodiments shown and described in the accompanying drawings are only exemplary and are intended to explain the principles and spirit of the present invention, rather than to limit the scope of the present invention.
[0059] like Figure 1 As shown, a single-camera fast spatial three-dimensional reconstruction measurement method includes the following steps:
[0060] S1, obtaining imaging distortion parameters of the camera probe;
[0061] The camera probe is a single-camera probe, and the camera probe is installed on a motion platform.
[0062] In this embodiment, step S1 specifically includes:
[0063] Use the calibration plate and the cross to calibrate the camera probe at different distances to obtain the imaging distortion parameters;
[0064] The method of using a calibration plate and a cross to calibrate the camera probe at different distances is the Zhang Zhengyou calibration method, or the industrial close-range photogrammetry calibration method.
[0065] Exemplarily, the motion platform can be a movable equipment such as a vehicle, a drone or an airplane. The camera probe can be placed at any position on the motion platform, but the relative relationship between the motion direction of the motion platform and the direction of the optical axis of the camera probe must be clear. The relative relationship between the motion direction of the motion platform and the direction of the optical axis of the camera probe can be achieved through camera calibration or installation positioning. In the process of camera calibration, calibration can be performed using the calibration of the calibration plate and the cross, and the calibration of the calibration plate and the cross can be performed based on Zhang Zhengyou's calibration method or the industrial close-range photogrammetry calibration method.
[0066] For example, a small camera can be installed on the UAV platform. The small camera should be firmly installed on the UAV platform to ensure that the small camera will not loosen during the flight of the UAV platform. For small cameras, a 2m long cross is used for calibration in a 40m spatial range. The calibration method can use the close-range photogrammetry calibration method to obtain the imaging distortion model parameters of the camera imaging system.
[0067] Optionally, in this embodiment, multiple camera probes can be placed on the motion platform at the same time, and each camera probe can independently image and be unified in the same coordinate system based on the coordinate system setting in S2, or multi-angle synchronous reconstruction can be performed in conjunction with multi-viewing vision.
[0068] S2. Setting a measurement coordinate system and acquiring continuous images captured by the camera probe in the measurement coordinate system.
[0069] In this embodiment, in step S2, setting the measurement coordinate system specifically includes:
[0070] S21. Set at least 5 marking points in the measurement space;
[0071] S22, using a camera probe to capture images of the landmark at three different motion positions;
[0072] S23, using the resection method to solve the external parameter difference of the camera probe according to the image of the marker point;
[0073] S24, obtaining a corresponding relationship between a motion coordinate system of the camera probe and an imaging coordinate system of the camera probe according to the difference in external parameters;
[0074] The formula used to obtain the correspondence between the motion coordinate system of the camera probe and the imaging coordinate system of the camera probe is:
[0075]
[0076] Among them, P f Indicates the spatial coordinates of the camera probe at the time of shooting, which can be obtained using GPS and other equipment. c The image of the camera probe is based on the coordinates of the optical center of the intersection solution of the marker point coordinates, R fc and T fc Represents the correspondence between the motion coordinate system of the camera probe and the imaging coordinate system of the camera probe;
[0077] S25. Setting a measurement coordinate system according to the correspondence between the motion coordinate system of the camera probe and the imaging coordinate system of the camera probe.
[0078] For example, the motion platform is a drone. The spatial relationship between the flight coordinate system of the drone and the camera imaging system can be calibrated as follows: in the measurement space, more than 5 marker points are used for arrangement, and the coordinates of the marker points can be measured using the method of photogrammetric reconstruction or total station reconstruction. For example, the drone can take images of the same marker point at three different motion positions, and use the rear intersection method to solve the relative spatial external parameter difference of the camera at different positions of the drone at the same time, compare the relative spatial external parameter difference with the motion interval of the drone flight parameter, and obtain the corresponding relationship between the motion coordinate system of the camera probe and the imaging coordinate system of the camera probe. For another example, the drone can be started, and after stabilization, the current drone coordinates can be recorded using GPS as the initial coordinates of the measurement. The starting position of the drone can also be used as the coordinate origin during spatial measurement, or the spatial coordinate system coordinates can be unified using a total station as the setting of the measurement coordinate system. After the measurement coordinate system is set, the image data of the drone during flight is recorded. Among them, the measurement coordinate system refers to the measurement coordinate system fixed during the entire measurement process, and the measurement coordinate system can be given based on the initial take-off position of the drone during the initial flight, or by a position giving means such as GPS. After the measurement coordinate system is given, it will not change during the entire measurement process.
[0079] S3. Extract features from continuous images to obtain feature key points and edge information.
[0080] In this embodiment, in step S3, feature extraction is performed on the continuous images, and the methods used include edge extraction algorithm, SIFT algorithm and SURF algorithm.
[0081] Exemplarily, SIFT can be used to identify target key points. For example, two images that are continuous in time sequence can be used as feature spectrograms for corresponding identification. In this way, a large number of two images that are continuous in time sequence can be used as identification data to establish a feature key point data set, and as time increases, the data in the feature key point data set will continue to increase. When the size of the feature key point data set exceeds a certain size, the feature key point data set can be updated by a forgetting algorithm to maintain the recent validity of the feature key point data set. For the continuously updated feature key point data set, feature identification and matching can be performed in the latest updated image, and the new features can be updated into the feature key point data set. In addition, for the startup moment of the motion platform, a feature key point data set can be established based on the first image captured by the camera probe installed on the motion platform.
[0082] S4. Reconstruct the target object in the continuous image according to the feature key points and the imaging distortion parameters to obtain the three-dimensional coordinates of the feature key points, and segment the target object according to the edge information to obtain the edge segmentation result of the target object.
[0083] In this embodiment, in step S4, the target object in the continuous image is reconstructed according to the feature key points and the imaging distortion parameters, and the target object reconstruction formula used is:
[0084] A=[P i ' -1 -C i-1 C i -P i ′]
[0085] B=[P i ′-P i ' -1 ]
[0086] A*P i 3D =B
[0087] Among them, A and B represent parameter matrices, i represents the image sequence index number, P i 3D represents the three-dimensional coordinates of the feature key points in the i-th image, P i ′ represents the three-dimensional coordinates of the key feature points in the i-th image on the imaging plane after removing the imaging distortion, P i ' -1 represents the three-dimensional coordinates of the key points in the i-1th image on the imaging plane after removing the imaging distortion, C i represents the optical center coordinates of the camera in the i-th image, C i-1 Represents the coordinates of the optical center of the camera in the i-1th image.
[0088] In this embodiment, the method for solving the three-dimensional coordinates of the feature key points on the imaging plane after removing the imaging distortion may be the Zhang Zhengyou distortion model;
[0089] Zhang Zhengyou's distortion model is:
[0090]
[0091] in, and represents the two-dimensional coordinates of the imaging plane after the imaging distortion is removed from the feature key points, x and y represent the actual two-dimensional coordinates of the feature key points extracted from the acquired image, r represents the distance from the imaging point of the feature key point in the two-dimensional image to the imaging center, u0 and v0 represent the coordinates of the imaging optical center of the camera probe obtained by calibration in the imaging plane, k1 and k2 represent the imaging distortion parameters obtained using Zhang Zhengyou's calibration method.
[0092] In this embodiment, the method for solving the three-dimensional coordinates of the feature key points on the imaging plane after removing the imaging distortion can also be an industrial photogrammetry model;
[0093] The industrial photogrammetry model is:
[0094]
[0095] in, and represents the two-dimensional coordinates of the feature key point in the imaging plane after the imaging distortion is removed, x and y represent the actual two-dimensional coordinates of the feature key point extracted from the acquired image, r represents the distance between the imaging point of the feature key point in the two-dimensional image and the imaging center, u0 and v0 represent the coordinates of the imaging optical center of the camera probe obtained by calibration in the imaging plane, A1, A2, A3, B1, B2, C1 and C2 represent the imaging distortion parameters obtained using the industrial photogrammetry calibration method.
[0096] In this embodiment, in step S4, the target object in the continuous image is reconstructed according to the feature key points and the imaging distortion parameters to obtain the three-dimensional coordinates of the feature key points, which specifically includes:
[0097] S41, obtaining the coordinates of the two-dimensional points on the imaging plane after removing the imaging distortion according to the characteristic key points and the imaging distortion parameters;
[0098] S42, according to a conversion formula, converting the coordinates of the two-dimensional point on the imaging plane after removing the imaging distortion into the three-dimensional coordinates of the feature key point on the imaging plane after removing the imaging distortion;
[0099] The conversion formula is:
[0100]
[0101] Where P′ represents the three-dimensional coordinates of the key points on the imaging plane after removing the imaging distortion. and represents the two-dimensional coordinates of the imaging plane after removing the imaging distortion of the feature key points, K pz is the pixel size corresponding to the pixel, that is, the coefficient corresponding to the physical size of the image pixel on the optical sensor, R cam The relationship between the optical center, optical axis and imaging optical sensor in the imaging optical path, in the imaging system of ordinary industrial cameras The positive and negative signs in the ±f term are determined according to the definition of the optical center position, and f represents the focal length, that is, the actual physical size focal length of the imaging system;
[0102] S43, obtaining the three-dimensional coordinates of the feature key points according to the three-dimensional coordinates of the feature key points on the imaging plane after removing the imaging distortion.
[0103] In this embodiment, in step S4, the method used to segment the target object is edge extraction based on digital image processing, or feature extraction, segmentation and recognition based on artificial intelligence.
[0104] For example, in this embodiment, an industrial camera with a focal length of 8mm and a pixel size of 4.4um is used as an example. Three-dimensional reconstruction of key points is performed based on the feature points in two consecutive images in time sequence, and based on the single camera imaging method, the system known quantity is the known parameters of the camera probe imaging, and the camera probe imaging distortion parameters can be calibrated and obtained in step S1. The three-dimensional coordinates of the feature key points can be obtained by solving the target object reconstruction formula.
[0105] Optionally, in the target object reconstruction formula, taking into account the errors of the optical path imaging system in real image recognition and measurement, in order to obtain a more accurate solution, instead of using the traditional linear solution method, an optimization method such as the least squares method, SVD decomposition and other optimization methods can be used to solve the three-dimensional coordinates of the feature key points.
[0106] Exemplarily, an edge extraction algorithm is used to obtain edges and perform feature segmentation. In this embodiment, canny detection is used to perform edge detection of time-series continuous images, and feature extraction, segmentation and recognition are performed based on the edge continuity of feature key points in two time-series continuous images.
[0107] S5. Perform three-dimensional space reconstruction according to the three-dimensional coordinates of the feature key points and the edge segmentation results to obtain a three-dimensional space reconstruction result.
[0108] In this embodiment, step S5 specifically includes:
[0109] Determine whether the three-dimensional coordinates of the key feature points meet the point static evaluation conditions;
[0110] When the three-dimensional coordinates of the feature key points meet the point static evaluation conditions, three-dimensional space reconstruction is performed according to the three-dimensional coordinates of the feature key points and the edge segmentation results to obtain the three-dimensional space reconstruction results;
[0111] When the three-dimensional coordinates of the feature key points do not meet the point static evaluation conditions, reselect the feature key points;
[0112] The static evaluation conditions of the point are:
[0113]
[0114] Among them, P i 3D Represents the three-dimensional coordinates of the feature key points in the i-th image, Represents the three-dimensional coordinates of the key feature points in the i-1th image, Threshoud Static Represents the spatial feature stability threshold.
[0115] Exemplarily, step S5 can be used in conjunction with binoculars to improve the accuracy of three-dimensional reconstruction of fixed features.
[0116] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.
Claims
1. A single-camera fast spatial three-dimensional reconstruction measurement method, characterized in that: The method comprises: S1, obtaining imaging distortion parameters of the camera probe; Wherein, the camera probe is a single-camera probe, and the camera probe is installed on a motion platform; S2, setting a measurement coordinate system, and acquiring continuous images captured by the camera probe in the measurement coordinate system; S3, extracting features from the continuous images to obtain feature key points and edge information; S4, reconstructing the target object in the continuous image according to the feature key points and the imaging distortion parameters to obtain the three-dimensional coordinates of the feature key points, and segmenting the target object according to the edge information to obtain an edge segmentation result of the target object; S5. Perform three-dimensional space reconstruction according to the three-dimensional coordinates of the feature key points and the edge segmentation result to obtain a three-dimensional space reconstruction result.
2. The method according to claim 1, characterized in that The step S1 specifically includes: Using a calibration plate and a cross to calibrate the camera probe at different distances to obtain the imaging distortion parameters; The method of using a calibration plate and a cross to calibrate the camera probe at different distances is the Zhang Zhengyou calibration method, or the industrial close-range photogrammetry calibration method.
3. The method according to claim 1, characterized in that In step S2, setting the measurement coordinate system specifically includes: S21. Set at least 5 marking points in the measurement space; S22, using the camera probe to capture images of the marker points at three different movement positions; S23, solving the external parameter difference of the camera probe using the resection method according to the image of the marker point; S24, acquiring a corresponding relationship between a motion coordinate system of the camera probe and an imaging coordinate system of the camera probe according to the difference in the external parameters; The formula used to obtain the correspondence between the motion coordinate system of the camera probe and the imaging coordinate system of the camera probe is: Among them, P f represents the spatial coordinates of the camera probe at the shooting time, P c The image of the camera probe is based on the coordinates of the optical center of the intersection solution of the marker point coordinates, R fc and T fc Represents the correspondence between the motion coordinate system of the camera probe and the imaging coordinate system of the camera probe; S25. Setting the measurement coordinate system according to the correspondence between the motion coordinate system of the camera probe and the imaging coordinate system of the camera probe.
4. The method according to claim 1, characterized in that: In the step S3, the method of extracting features from the continuous images includes edge extraction algorithm, SIFT algorithm and SURF algorithm.
5. The method according to claim 1, characterized in that In step S4, the target object in the continuous image is reconstructed according to the feature key points and the imaging distortion parameters, and the target object reconstruction formula used is: A=[P i ′ -1 -C i-1 C i -P i ′] B=[P i ′-P i ′ -1 ] A*P i 3D =B Among them, A and B represent parameter matrices, i represents the image sequence index number, P i 3D represents the three-dimensional coordinates of the feature key points in the i-th image, P i ′ represents the three-dimensional coordinates of the key feature points in the i-th image on the imaging plane after removing the imaging distortion, P i ' -1 represents the three-dimensional coordinates of the key points in the i-1th image on the imaging plane after removing the imaging distortion, C i represents the optical center coordinates of the camera in the i-th image, C i-1 Represents the coordinates of the optical center of the camera in the i-1th image.
6. The method according to claim 5, characterized in that The method for solving the three-dimensional coordinates of the characteristic key points on the imaging plane after removing the imaging distortion is the Zhang Zhengyou distortion model; The Zhang Zhengyou distortion model is: in, and represents the two-dimensional coordinates of the imaging plane after the imaging distortion is removed from the feature key points, x and y represent the actual two-dimensional coordinates of the feature key points extracted from the acquired image, r represents the distance from the imaging point of the feature key point in the two-dimensional image to the imaging center, u0 and v0 represent the coordinates of the imaging optical center of the camera probe obtained by calibration in the imaging plane, k1 and k2 represent the imaging distortion parameters obtained using Zhang Zhengyou's calibration method.
7. The method according to claim 5, characterized in that The method for solving the three-dimensional coordinates of the characteristic key points on the imaging plane after removing the imaging distortion is an industrial photogrammetry model; The industrial photogrammetry model is: in, and represents the two-dimensional coordinates of the feature key point in the imaging plane after the imaging distortion is removed, x and y represent the actual two-dimensional coordinates of the feature key point extracted from the acquired image, r represents the distance between the imaging point of the feature key point in the two-dimensional image and the imaging center, u0 and v0 represent the coordinates of the imaging optical center of the camera probe obtained by calibration in the imaging plane, A1, A2, A3, B1, B2, C1 and C2 represent the imaging distortion parameters obtained using the industrial photogrammetry calibration method.
8. The method according to claim 6 or 7, characterized in that: In the step S4, the target object in the continuous image is reconstructed according to the feature key points and the imaging distortion parameters to obtain the three-dimensional coordinates of the feature key points, which specifically includes: S41, acquiring the coordinates of the two-dimensional points on the imaging plane after removing the imaging distortion according to the characteristic key points and the imaging distortion parameters; S42, according to a conversion formula, converting the two-dimensional coordinates of the imaging plane after removing the imaging distortion into the three-dimensional coordinates of the feature key point on the imaging plane after removing the imaging distortion; The conversion formula is: Where P′ represents the three-dimensional coordinates of the key points on the imaging plane after removing the imaging distortion. and represents the two-dimensional coordinates of the imaging plane after removing the imaging distortion of the feature key points, K pz is the pixel size corresponding to the pixel, that is, the coefficient corresponding to the physical size of the image pixel on the optical sensor, R cam The relationship between the optical center, optical axis and imaging optical sensor in the imaging optical path, in the imaging system of ordinary industrial cameras The positive and negative signs in the ±f term are determined according to the definition of the optical center position, and f represents the focal length, that is, the actual physical size focal length of the imaging system; S43, acquiring the three-dimensional coordinates of the feature key point according to the three-dimensional coordinates of the feature key point on the imaging plane after removing imaging distortion.
9. The method according to claim 1, characterized in that: In step S4, the method used for segmenting the target object is edge extraction based on digital image processing, or feature extraction, segmentation and recognition based on artificial intelligence.
10. The method according to claim 5, characterized in that The step S5 specifically includes: Determine whether the three-dimensional coordinates of the key feature points meet the point static evaluation conditions; When the three-dimensional coordinates of the feature key points meet the point static evaluation conditions, three-dimensional space reconstruction is performed according to the three-dimensional coordinates of the feature key points and the edge segmentation result to obtain a three-dimensional space reconstruction result; When the three-dimensional coordinates of the feature key points do not meet the point static evaluation conditions, reselect the feature key points; The point static evaluation conditions are: Among them, P i 3D Represents the three-dimensional coordinates of the feature key points in the i-th image, Represents the three-dimensional coordinates of the key feature points in the i-1th image, Threshoud Static Represents the spatial feature stability threshold.
