Large-field-of-view three-dimensional fusion two-dimensional visual inspection system calibration method

Abstract

The invention relates to a calibration method for a large-view-field three-dimensional fusion two-dimensional visual detection system, which comprises the following steps of: calibrating parameters of the large-view-field three-dimensional fusion two-dimensional visual detection system based on a visual calibration target, a standard ball bat and a laser tracking three-dimensional measurement system, and calibrating internal and external parameters of a camera and a light source projector through the visual calibration target; calibration of a large-view-field splicing matrix is performed through a laser tracking three-dimensional measurement system, and calibration results can meet three-dimensional and two-dimensional detection requirements of a visual detection system at the same time, so that full-range misloading and neglected loading detection of the honeycomb sandwich structure plate of the spacecraft based on point cloud comparison and contour comparison is realized, and detection requirements of different measurement objects are met.

Description

Technical Field

[0001] The invention relates to a calibration method for a large-field-of-view three-dimensional fusion two-dimensional visual detection system, and belongs to the technical field of large-field-of-view visual detection. Background Art

[0002] The large-field-of-view 3D fusion 2D visual inspection system is used for the inspection of honeycomb sandwich structure panels of spacecraft. Honeycomb sandwich structure panels are composed of face sheets, structural adhesives, inserts and sandwich honeycomb cores. Before the upper face sheet is closed, the visual inspection system is required to inspect the assembly conformity of the inserts, including embedded parts, honeycomb core encryption and L direction, foam reinforcement area, etc.

[0003] In view of the characteristics of the spacecraft honeycomb sandwich structure inspection task, a large-field 3D fusion 2D visual inspection system composed of multiple cameras and light source projectors was developed. The large-scale 3D inspection technology was used to obtain the full-range 3D point cloud of the object under inspection. Multiple 2D images of the object under inspection were obtained by 2D inspection. Through 3D point clouds and 2D images, based on 3D fusion 2D inspection technology, the full-range mis-assembly inspection of spacecraft honeycomb sandwich structure panels was realized based on point cloud comparison and contour comparison.

[0004] The large-field-of-view 3D fusion 2D visual inspection system consists of multiple subsystems. Before use, the internal and external parameters of each subsystem component and the large-field-of-view stitching matrix calibration need to be performed. The purpose of the internal and external parameter calibration is to determine the mapping relationship between the points in the image and the actual space points. The internal parameters include the focal length of the camera, the coordinates of the principal point, etc.; the external parameters include the rotation matrix and translation vector of the camera, which are used to describe the posture and position of the camera in space.

[0005] At present, the calibration method of visual inspection system is mainly aimed at single camera or binocular system calibration. The calibration method is simple and universal, suitable for standard visual inspection system calibration, but cannot be applied to the calibration of large field of view three-dimensional fusion two-dimensional visual inspection system. Summary of the invention

[0006] The technical problem solved by the present invention is: in view of the fact that the traditional visual inspection system calibration method in the current existing technology is not suitable for the calibration requirements of the current large-field-of-view three-dimensional fusion two-dimensional visual inspection system, a large-field-of-view three-dimensional fusion two-dimensional visual inspection system calibration method is proposed.

[0007] The present invention solves the above technical problems by the following technical solutions:

[0008] A calibration method for a large-field-of-view three-dimensional fusion two-dimensional visual inspection system, comprising:

[0009] Adjust the camera lens and light projector lens in the visual inspection system;

[0010] Place visual targets at the four corners and the center of each camera's field of view, and calibrate the internal parameters of each camera based on the images taken at different positions of the visual targets.

[0011] Place visual targets in the common field of view of adjacent cameras, and calibrate the external parameters of each camera based on the images of each visual target at different positions and the positions of each visual target relative to each camera.

[0012] Place visual calibration targets within the projection area of ​​each light source projector, and calibrate the internal and external parameters of each light source projector according to the imaging images of each visual target in different positions;

[0013] A standard golf club is placed within the projection area of ​​each light source projector, and a spherical point cloud of the standard golf club is obtained after imaging the standard golf club. The spherical center distance is obtained by fitting processing, and the translation vector of the light source projector relative to the camera is corrected by the measured values ​​of each group of spherical center distances;

[0014] A measuring target ball is placed within the full field of view of the visual inspection system, and the coordinate set of the center of the ball is obtained using the visual inspection system. After fitting processing, the large field of view stitching matrix of the visual inspection system is obtained to complete the calibration of the visual inspection system.

[0015] The adjustment constraints of the camera lens and light source projector lens in the visual inspection system are:

[0016] Adjust the focal length of the camera lens so that the edges of the visual targets at the four corners and the center of the camera's field of view are clear;

[0017] Adjust the light source projector lens to optimize the clarity of the fine lines projected at the four corners of the frame.

[0018] The method for calibrating the internal parameters of each camera is:

[0019] After placing visual targets at m positions, including the four corners and the center, within the field of view of each camera, n types of poses are set for the visual targets at each position, where n is a positive integer. m×n images of each camera are determined, and the target center of each image is extracted. Based on the known spatial positions of the visual targets of each camera and the two-dimensional pixel points of the imaging image, the internal parameters of each camera are calculated by minimizing the reprojection error.

[0020] The method for calibrating the external parameters of each camera is:

[0021] After placing visual targets at m positions in the common field of view of each adjacent camera, n types of postures are set for the visual targets at each position, where n is a positive integer. The number of m×n images imaged by each camera is determined, and the geometric relationship of adjacent camera groups in the world coordinate system is determined according to the position of camera i in the world coordinate system. After traversing all adjacent camera groups, the geometric relationships of all adjacent camera groups are used as external parameters and unified to the coordinate system of the first camera.

[0022] The internal parameter calibration method of each light source projector is as follows:

[0023] Visual calibration targets are placed at m positions within the projection area of ​​each light source projector. N types of postures are set for the visual targets at each position, where n is a positive integer. When the light source projector projects stripes, the camera takes photos synchronously to obtain m×n groups of imaging images. Each group of images contains stripe images with different frequencies and a phase difference of π / 2.

[0024] The absolute phase values ​​of the horizontal grating image and the vertical grating image are calculated by using the four-step phase shift method and the heterodyne principle, and the absolute phase values ​​corresponding to the center coordinates of each circle in the X and Y directions are obtained by using linear interpolation. The center coordinates of each circle on the target are calibrated in the image coordinate system of the light source projector using the absolute phase values ​​to obtain m×n sets of center coordinates.

[0025] Using the m×n sets of circle center coordinates, the internal parameters of the light projector are calculated by minimizing the reprojection error.

[0026] The method for determining the external parameters of each light source projector is as follows:

[0027] According to the internal parameters of each camera, the rotation and translation matrix between the coordinate systems of each camera and the light source projector is obtained, and the external parameters of each light source projector are obtained after conversion using the internal parameters of each light source projector and the rotation and translation matrix.

[0028] The calculation method of the correction amount of the translation vector of the light source projector relative to the camera is:

[0029] Standard bats are placed at m positions within the projection format of each light projector. When the light projector projects stripes, the camera synchronously captures images to obtain the spherical point clouds of the standard balls at both ends of the standard bats. The coordinates of the two sphere centers are obtained by fitting the sphere point clouds to the spheres. The distance between the two sphere centers of the bats is obtained, and the translation vector of the light projector relative to the camera is corrected using multiple sets of measured values ​​and nominal values ​​of the sphere center distances.

[0030] The corrected translation vector t' of the light source projector relative to the camera is:

[0031]

[0032]

[0033] In the formula, is the average value of the ball center distance measurement; n is the number of club positions; l i is the ith measurement value of the sphere center distance; l r is the nominal value of the sphere center distance; t is the calibrated translation vector of the light source projector relative to the camera.

[0034] The method for obtaining the large field of view stitching matrix of the visual inspection system is as follows:

[0035] Place n measuring target balls within the full field of view of the visual inspection system;

[0036] The subsystem in the visual inspection system is used to measure the three-dimensional point cloud of each measurement target ball, which is used to obtain the spherical center coordinate set P of each measurement target ball. 视1 , P 视2 , P 视3 …P 视n ;

[0037] The overall spherical center coordinate set P is obtained through finite point measurement and fitting processing of the laser tracking 3D measurement system. 激 ;

[0038] The coordinate sets of each sphere center and the overall sphere center coordinate set P are respectively 激 The fitting process is performed to obtain a large field of view stitching matrix.

[0039] The large field of view stitching matrix includes a rotation and translation matrix from the visual detection subsystem coordinate system to the laser tracking three-dimensional measurement system coordinate system, and a rotation and translation matrix between adjacent visual detection subsystems;

[0040] The method for obtaining the rotation and translation matrix from the visual inspection subsystem coordinate system to the laser tracking 3D measurement system coordinate system is:

[0041] The laser tracking 3D measurement system is in its own coordinate system F 激 The target ball center point set measured below is P 激 {p 激1 , p 激2 , p 激3 ,…p 激n-1 , p 激n}, detection subsystem i in its own coordinate system F i The target ball center point set measured below is P i {p i1 , p i2 , p i3 , p i4 , p i5 , p i6 ...}, the coordinate system F of the visual detection subsystem i i and the coordinate system F of the visual inspection subsystem i+1 i+1 To the laser tracking 3D measurement system coordinate system F 激 The rotation and translation matrices are R 激i , T 激i and R 激i+1 , T 激i+1 ,in:

[0042] P激 =R 激i P i +T 激i

[0043] P 激 =R 激i+1 P i+1 +T 激i+1

[0044] The method for obtaining the rotation and translation matrix between adjacent visual detection subsystems is:

[0045] R 视i+1 =R 激i+1 R 激i -1

[0046] T 视i+1 =T 激i+1 -R 激i+1 R 激i -1 T 激i .

[0047] In the formula, the rotation and translation matrix R 激i , T 激i and R 激i+1 , T 激i+1 Obtained through singular value decomposition.

[0048] The advantages of the present invention compared with the prior art are:

[0049] (1) The present invention provides a calibration method for a large-field-of-view 3D fusion 2D visual inspection system. The large-field-of-view 3D fusion 2D visual inspection system based on a visual calibration target, a standard ball stick and a laser tracking 3D measurement system realizes parameter calibration. The internal and external parameters of the camera and the light source projector are calibrated by the visual calibration target. The large-field-of-view stitching matrix is ​​calibrated by the laser tracking 3D measurement system. The visual calibration target and the standard ball stick are used as measurement objects. The light source projector in the system projects stripes and the camera takes images separately or synchronously to obtain the internal and external parameters of the camera and the light source projector. The calibration result can simultaneously meet the 3D and 2D detection requirements of the visual inspection system, thereby realizing the full range of mis-installation detection of honeycomb sandwich structure panels of spacecraft based on point cloud comparison and contour comparison, and meeting the detection requirements of different measurement objects.

[0050] (2) The visual inspection system calibration method of the present invention includes a large field of view stitching matrix calibration method, which uses the laser tracking three-dimensional measurement system measurement point set as the reference point set, and obtains the transformation matrix of different fields of view through the best fit of the measurement point set and the reference point set under the corresponding fields of view of different visual inspection subsystems, thereby realizing large field of view data stitching. The laser tracking three-dimensional measurement system has a large measurement range, high measurement accuracy, and good portability, making the calibration method more applicable. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A schematic diagram of the camera intrinsic parameter calibration provided by the present invention;

[0052] Figure 2 A schematic diagram of adjacent camera calibration provided by the present invention;

[0053] Figure 3 A schematic diagram of the calibration of the internal and external parameters of the light source projector provided by the present invention;

[0054] Figure 4 A schematic diagram of length tracing provided by the present invention;

[0055] Figure 5 This is a schematic diagram of the large field of view stitching matrix calibration provided by the present invention. DETAILED DESCRIPTION

[0056] A calibration method for a large-field-of-view 3D-fused 2D visual inspection system is disclosed. The method performs parameter calibration on the large-field-of-view 3D-fused 2D visual inspection system based on a visual calibration target, a standard ball stick and a laser tracking 3D measurement system. The internal and external parameters of a camera and a light source projector are calibrated by the visual calibration target. The large-field-of-view stitching matrix is ​​calibrated by the laser tracking 3D measurement system. The calibration result can simultaneously meet the 3D and 2D inspection requirements of the visual inspection system, thereby realizing a full range of mis-installation and missing-installation inspection of honeycomb sandwich structure panels of spacecraft based on point cloud comparison and contour comparison, and meeting the inspection requirements of different measurement objects.

[0057] The calibration method of the large field of view 3D fusion 2D visual inspection system includes the following steps:

[0058] Adjust the camera lens and light projector lens in the visual inspection system;

[0059] Place visual targets at the four corners and the center of each camera's field of view, and calibrate the internal parameters of each camera using images taken at different positions of each visual target.

[0060] Place visual targets in the common field of view of adjacent cameras, and determine the external parameters of each camera based on the images of each visual target at different positions and the positions of each visual target relative to each camera;

[0061] Place visual calibration targets within the projection area of ​​each light source projector, and calibrate the internal and external parameters of each light source projector based on the imaging images of each visual target at different positions;

[0062] A standard golf club is placed within the projection area of ​​each light source projector, and a point cloud of the standard golf club spherical surface is obtained. The spherical center distance is obtained by fitting processing, and the translation vector of the light source projector relative to the camera is corrected by the measured value and nominal value of each group of spherical center distances;

[0063] A measurement target sphere is placed within the full field of view of the visual inspection system to obtain a set of sphere center coordinates. After fitting processing, the large field of view stitching matrix of the visual inspection system is obtained to complete the calibration of the visual inspection system.

[0064] The adjustment constraints of the camera lens and light source projector lens in the visual inspection system are:

[0065] Adjust the focal length of the camera lens so that the edges of the visual targets at the four corners and the center of the camera's field of view are clear;

[0066] Adjust the light source projector lens to optimize the clarity of the fine lines projected at the four corners of the frame.

[0067] After placing visual targets at m positions, including the four corners and the center, within the field of view of each camera, n types of poses are set for the visual targets at each position, where n is a positive integer. m×n images of each camera are determined, and the target center of each image is extracted. Based on the known spatial positions of the visual targets of each camera and the two-dimensional pixel points of the imaging image, the internal parameters of each camera are calculated by minimizing the reprojection error.

[0068] After placing visual targets at m positions in the common field of view of each adjacent camera, n types of postures are set for the visual targets at each position, where n is a positive integer. The number of m×n images imaged by each camera is determined, and the geometric relationship of adjacent camera groups in the world coordinate system is determined according to the position of camera i in the world coordinate system. After traversing all adjacent camera groups, the geometric relationships of all adjacent camera groups are used as external parameters and unified to the coordinate system of the first camera.

[0069] The internal parameter calibration method of each light source projector is as follows:

[0070] Visual calibration targets are placed at m positions within the projection area of ​​each light source projector. N types of postures are set for the visual targets at each position, where n is a positive integer. When the light source projector projects stripes, the camera takes photos synchronously to obtain m×n groups of imaging images. Each group of images contains stripe images with different frequencies and a phase difference of π / 2.

[0071] The absolute phase values ​​of the horizontal grating image and the vertical grating image are calculated by using the four-step phase shift method and the heterodyne principle, and the absolute phase values ​​corresponding to the center coordinates of each circle in the X and Y directions are obtained by linear interpolation. The center coordinates of each circle on the calibration target in the image coordinate system of the light source projector are calculated by using the absolute phase values.

[0072] Using the m×n sets of circle center coordinates, the internal parameters of the light projector are calculated by minimizing the reprojection error.

[0073] The external parameters of each light source projector are determined as follows:

[0074] According to the internal parameters of each camera, the rotation and translation matrix between the coordinate systems of each camera and the light source projector is obtained, that is, the external parameters of each light source projector are obtained.

[0075] The correction value of the light projector's translation vector relative to the camera is calculated as:

[0076] Standard bats are placed at m positions within the projection format of each light source projector. When the light source projector projects stripes, the camera synchronously captures images to obtain the spherical point clouds of the standard balls at both ends of the standard bats. The coordinates of the two sphere centers are obtained by fitting the sphere point clouds to the spheres. The distance between the two sphere centers of the bats is obtained, and the translation vector of the light source projector relative to the camera is corrected using the measured values ​​and nominal values ​​of each set of sphere center distances.

[0077] The method for obtaining the large field of view stitching matrix of the visual inspection system is:

[0078] Place n measuring target balls within the full field of view of the visual inspection system;

[0079] The subsystem in the visual inspection system is used to measure the three-dimensional point cloud of each measurement target ball, which is used to obtain the spherical center coordinate set P of each measurement target ball. 视1 , P 视2 , P 视3 …P 视n ;

[0080] The overall spherical center coordinate set P is obtained through finite point measurement and fitting processing of the laser tracking 3D measurement system. 激 ;

[0081] The coordinate sets of each sphere center and the overall sphere center coordinate set P are respectively 激 The fitting process is performed to obtain a large field of view stitching matrix.

[0082] The following is further described in conjunction with the accompanying drawings and preferred embodiments of the specification:

[0083] In the current embodiment, the camera intrinsic parameters are calibrated as follows: Figure 1 As shown in the figure, in the calibration method, for camera intrinsic calibration, visual targets are placed at m positions in each camera field of view, each position is transformed into n poses, and each camera intrinsic calibration takes a total of m×n images. The target center of each image is extracted, and the internal parameters of the camera are calculated by minimizing the reprojection error using these known three-dimensional spatial points and their corresponding two-dimensional pixel points in the image.

[0084] Adjacent cameras are calibrated as Figure 2As shown in the figure, taking four cameras as an example, there is a common field of view between the four cameras. Taking camera 0 as the reference, visual calibration targets are placed at m positions in the common field of view of camera 0 and camera 1, and each position is transformed into n poses. Let the external parameters of camera 0 and camera 1 be R0, T0 and R1, T1 respectively, then R0, T0 represent the relative position of camera 0 and the world coordinate system, and R1, T1 represent the relative position of camera 1 and the world coordinate system. The geometric relationship between the two cameras R 01 , T 01 It can be expressed as:

[0085] R 01 =R1R0 -1

[0086] T 01 =T1-R1 R0 -1 T0

[0087] Similarly, the geometric relationship R between camera 0 and camera 2, camera 0 and camera 3 can be calculated 02 , T 02 and R 03 , T 03 , the coordinate systems of the four cameras can be converted to the coordinate system of camera 0.

[0088] The internal and external parameters of the light source projector are calibrated as follows: Figure 3 As shown in the figure, visual calibration targets are placed at m positions within the projection area of ​​each light projector, and each position changes n positions. When the light projector projects stripes, the camera shoots synchronously, and a total of m×n groups of images are shot. Each group of images contains stripe patterns with different frequencies and a phase difference of π / 2. The absolute phase values ​​of the horizontal and vertical grating images are calculated respectively by the four-step phase shift method and the heterodyne principle, and the absolute phase values ​​corresponding to each circle center coordinate in the X and Y directions are obtained by linear interpolation. Then, the X and Y coordinates of the corresponding digital micromirror (DMD) image are calculated by the absolute phase values, so as to obtain each circle center coordinate on the calibration target in the image coordinate system of the light projector. By processing each group of images, m×n groups of circle center coordinate sets can be obtained. Using these known three-dimensional space points and their corresponding two-dimensional pixel points in the image, the internal parameters of the light projector are calculated by minimizing the reprojection error. At the same time, the rotation and translation matrix between the coordinate systems of the camera and the light projector can be obtained according to the camera internal parameters, that is, the external parameters of the light projector are obtained.

[0089] Length traceability Figure 4As shown in the figure, standard ball sticks are placed at m positions within the projection area of ​​each light source projector. When the light source projector in the corresponding visual detection subsystem projects stripes, the camera synchronously captures the image, obtains the spherical point cloud of the standard balls at both ends of the standard ball stick, and fits the spheres to obtain the two sphere center coordinates, thereby obtaining the sphere center distance. The translation vector of the light source projector relative to the camera can be corrected through multiple sets of measured values ​​of the sphere center distance. The corrected translation vector t' of the light source projector relative to the camera is:

[0090]

[0091]

[0092] In the formula, is the average value of the distance from the center of the ball;

[0093] n is the number of club positions;

[0094] l i is the ith sphere center distance measurement value;

[0095] l r is the nominal value of the sphere center distance;

[0096] t is the calibrated translation vector of the light source projector relative to the camera;

[0097] Large field of view stitching matrix calibration: Figure 5 As shown in FIG. 1 , the measurement target balls are placed at multiple positions within the full field of view of the large field of view three-dimensional fusion two-dimensional visual inspection system. The subsystem in the visual inspection system measures the three-dimensional point cloud of the target balls at the corresponding positions and obtains the sphere center coordinate set P 视1 , P 视2 , P 视3 …P 视n The laser tracking 3D measurement system obtains the spherical center coordinate set P through finite point measurement and fitting. 激 By P 视1 , P 视2 , P 视3 …P 视n With P 激 The best fit of , and finally obtain the splicing matrix. Take 6 sets of visual inspection subsystems as an example, the subsystems are numbered 0-5, and the transformation matrix of subsystems 1-5 relative to subsystem 0 is calculated. Assume that the laser tracking 3D measurement system is in its own coordinate system F 激 The target ball center point set measured below is P 激 {p 激1 , p 激2 , p 激3 ,…p 激41 , p 激42}, the target ball center point set measured by detection subsystem 0 in its own coordinate system F0 is P0{p01 , p 02 , p 03 , p 04 , p 05 , p 06 , p 07}, the target ball center point set measured by detection subsystem 1 in its own coordinate system F1 is P1{p 11 , p 12 , p 13 , p 14 , p 15 , p 16 , p 17}, let the coordinate system F0 of visual inspection subsystem 0 and the coordinate system F1 of visual inspection subsystem 1 be the coordinate system F of laser tracking 3D measurement system 激 The rotation and translation matrices are R 激0 , T 激0 and R 激1 , T 激1 ,but:

[0098] P 激 =R 激0 P0+T 激0

[0099] P 激 =R 激1 P1+T 激1

[0100] R can be obtained by singular value decomposition (SVD) 激0 , T 激0 and R 激1 , T 激1 The rotation and translation matrix R from the coordinate system F1 of visual detection subsystem 1 to the coordinate system F0 of visual detection subsystem 0 is 视01 , T 视01 They are:

[0101] R 视01 =R 激1 R 激0 -1

[0102] T 视01 =T 激1 -R 激1 R 激0 -1 T 激0

[0103] Similarly, R can be calculated 视02 , T 视02 , R 视03 , T 视03 , R 视04 , T 视04, R 视05 , T 视05 , R 视06 , T 视06 .

[0104] Although the present invention has been disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art may make possible changes and modifications to the technical solution of the present invention by using the methods and technical contents disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the protection scope of the technical solution of the present invention.

[0105] The contents not described in detail in the specification of the present invention belong to the common knowledge of those skilled in the art.

Claims

1. A calibration method for a large field of view three-dimensional fusion two-dimensional visual inspection system, characterized in that include: Adjust the camera lens and light projector lens in the visual inspection system; Place visual targets at the four corners and the center of each camera's field of view, and calibrate the internal parameters of each camera based on the images taken at different positions of the visual targets. Place visual targets in the common field of view of adjacent cameras, and calibrate the external parameters of each camera based on the images of each visual target at different positions and the positions of each visual target relative to each camera. Place visual calibration targets within the projection area of ​​each light source projector, and calibrate the internal and external parameters of each light source projector according to the imaging images of each visual target in different positions; A standard golf club is placed within the projection area of ​​each light source projector, and a point cloud of the standard golf club spherical surface is obtained after imaging the standard golf club. The spherical center distance is obtained by fitting processing, and the translation vector of the light source projector relative to the camera is corrected by the measured values ​​of each group of spherical center distances; A measuring target sphere is placed within the full field of view of the visual inspection system, and the coordinate set of the sphere center is obtained using the visual inspection system. After fitting processing, the large field of view stitching matrix of the visual inspection system is obtained to complete the calibration of the visual inspection system.

2. According to claim 1, a calibration method for a large field of view three-dimensional fusion two-dimensional vision detection system is characterized by: The adjustment constraints of the camera lens and light source projector lens in the visual inspection system are: Adjust the focal length of the camera lens so that the edges of the visual targets at the four corners and the center of the camera's field of view are clear; Adjust the light source projector lens to optimize the clarity of the fine lines projected at the four corners of the frame.

3. The calibration method of a large field of view three-dimensional fusion two-dimensional vision detection system according to claim 1 is characterized in that: The method for calibrating the internal parameters of each camera is: After placing visual targets at m positions, including the four corners and the center, within the field of view of each camera, n types of poses are set for the visual targets at each position, where n is a positive integer. m×n images of each camera are determined, and the target center of each image is extracted. Based on the known spatial positions of the visual targets of each camera and the two-dimensional pixel points of the imaging image, the internal parameters of each camera are calculated by minimizing the reprojection error.

4. The method for calibrating a large field of view three-dimensional fusion two-dimensional vision inspection system according to claim 1, characterized in that: The method for calibrating the external parameters of each camera is: After placing visual targets at m positions in the common field of view of each adjacent camera, n types of postures are set for the visual targets at each position, where n is a positive integer. The number of m×n images imaged by each camera is determined, and the geometric relationship of adjacent camera groups in the world coordinate system is determined according to the position of camera i in the world coordinate system. After traversing all adjacent camera groups, the geometric relationships of all adjacent camera groups are used as external parameters and unified to the coordinate system of the first camera.

5. The method for calibrating a large field of view three-dimensional fusion two-dimensional vision inspection system according to claim 1, characterized in that: The internal parameter calibration method of each light source projector is as follows: Visual calibration targets are placed at m positions within the projection area of ​​each light source projector. N types of postures are set for the visual targets at each position, where n is a positive integer. When the light source projector projects stripes, the camera takes photos synchronously to obtain m×n groups of imaging images. Each group of images contains stripe images with different frequencies and a phase difference of π / 2. The absolute phase values ​​of the horizontal grating image and the vertical grating image are calculated by using the four-step phase shift method and the heterodyne principle, and the absolute phase values ​​corresponding to the center coordinates of each circle in the X and Y directions are obtained by using linear interpolation. The center coordinates of each circle on the target are calibrated in the image coordinate system of the light source projector using the absolute phase values ​​to obtain m×n sets of center coordinates. Using the m×n sets of circle center coordinates, the internal parameters of the light projector are calculated by minimizing the reprojection error.

6. The method for calibrating a large field of view three-dimensional fusion two-dimensional vision inspection system according to claim 5, characterized in that: The method for determining the external parameters of each light source projector is as follows: According to the internal parameters of each camera, the rotation and translation matrix between the coordinate systems of each camera and the light source projector is obtained, and the external parameters of each light source projector are obtained after conversion using the internal parameters of each light source projector and the rotation and translation matrix.

7. The method for calibrating a large field of view three-dimensional fusion two-dimensional vision inspection system according to claim 1, characterized in that: The calculation method of the correction amount of the translation vector of the light source projector relative to the camera is: Standard bats are placed at m positions within the projection format of each light projector. When the light projector projects stripes, the camera synchronously captures images to obtain the spherical point clouds of the standard balls at both ends of the standard bats. The coordinates of the two sphere centers are obtained by fitting the sphere point clouds to the spheres. The distance between the two sphere centers of the bats is obtained, and the translation vector of the light projector relative to the camera is corrected using multiple sets of measured values ​​and nominal values ​​of the sphere center distances.

8. The method for calibrating a large field of view three-dimensional fusion two-dimensional vision inspection system according to claim 7, characterized in that: The corrected translation vector t' of the light source projector relative to the camera is: In the formula, is the average value of the ball center distance measurement; n is the number of club positions; l i is the ith measurement value of the sphere center distance; l r is the nominal value of the sphere center distance; t is the calibrated translation vector of the light source projector relative to the camera.

9. The method for calibrating a large field of view three-dimensional fusion two-dimensional vision inspection system according to claim 1, characterized in that: The method for obtaining the large field of view stitching matrix of the visual inspection system is as follows: Place n measuring target balls within the full field of view of the visual inspection system; The subsystem in the visual inspection system is used to measure the three-dimensional point cloud of each measurement target ball, which is used to obtain the spherical center coordinate set P of each measurement target ball. 视1 , P 视2 , P 视3 …P 视n ; The overall spherical center coordinate set P is obtained through finite point measurement and fitting processing of the laser tracking 3D measurement system. 激 ; The coordinate sets of each sphere center and the overall sphere center coordinate set P are respectively 激 The fitting process is performed to obtain a large field of view stitching matrix.

10. The method for calibrating a large field of view three-dimensional fusion two-dimensional vision inspection system according to claim 1, characterized in that: The large field of view stitching matrix includes a rotation and translation matrix from the visual detection subsystem coordinate system to the laser tracking three-dimensional measurement system coordinate system, and a rotation and translation matrix between adjacent visual detection subsystems; The method for obtaining the rotation and translation matrix from the visual inspection subsystem coordinate system to the laser tracking 3D measurement system coordinate system is: The laser tracking 3D measurement system is in its own coordinate system F 激 The target ball center point set measured below is P 激 {p 激1 , p 激2 , p 激3 ,…p 激n-1 , p 激n }, detection subsystem i in its own coordinate system F i The target ball center point set measured below is P i {p i1 , p i2 , p i3 , p i4 , p i5 , p i6 ...}, the coordinate system F of the visual detection subsystem i i and the coordinate system F of the visual inspection subsystem i+1 i+1 To the laser tracking 3D measurement system coordinate system F 激 The rotation and translation matrices are R 激i 、T 激i and R 激i+1 、T 激i+1 ,in: P 激 =R 激i P i +T 激i P 激 =R 激i+1 P i+1 +T 激i+1 The method for obtaining the rotation and translation matrix between adjacent visual detection subsystems is: R 视i+1 =R 激i+1 R 激i -1 T 视i+1 =T 激i+1 -R 激i+1 R 激i -1 T 激i 。 In the formula, the rotation and translation matrix R 激i 、T 激i and R 激i+1 、T 激i+1 Obtained through singular value decomposition.

Images