A multi-view line laser scanning device based on binocular stereo vision and a method for panoramic measurement of complex surfaces

Abstract

The present invention discloses a multi-view line laser scanning system based on binocular stereo vision and a method for panoramic measurement of complex surfaces. The system comprises two cameras, a line laser, a rotating platform, a controller, a computer for data processing connecting the controller and the cameras, and two plane mirrors for collecting multi-view images. The two cameras are at the same height and have a horizontal distance difference. The line laser is fixed on a rotating bracket on the rotating platform. The system is simple and convenient to install, has low installation requirements, solves the problem of light crosstalk between the plane mirrors, and does not need to consider the influence of stray light of the collected image on three-dimensional measurement during the measurement process. In addition, the system has high measurement efficiency, low device cost, and high cost performance. Conventional panoramic three-dimensional measurement systems are expensive, complex to set up, and tedious to calibrate. The multi-view line laser scanning device proposed by the present invention uses plane mirrors to achieve multi-view imaging, has low device cost, and high cost performance.

Description

Technical Field

[0001] The present invention relates to the field of optical measurement, in particular to a multi-viewing angle line laser scanning device based on binocular stereo vision and a method for panoramic measurement of complex surfaces. Background Art

[0002] Optical 3D measurement technology, due to its high efficiency, non-destructive nature, and high flexibility, has found widespread application in medical diagnosis, cultural heritage preservation, reverse engineering, industrial manufacturing, and intelligent security. With the development of smart cities and smart industries, simply obtaining 3D information from a portion of a sample is no longer sufficient to fully meet diverse needs. Complete measurement and analysis of the sample is often required, leading to the rapid development of panoramic 3D measurement technology.

[0003] Conventional measuring devices cannot capture images containing all the surface contour information of the sample from a single perspective, so panoramic measurements of the sample to be measured are required from different perspectives. However, conventional panoramic surface three-dimensional measurement systems often require multiple sets of measuring devices or require multiple measurements, facing problems such as complex system setup, high cost, and high computational cost. Recently, another novel solution has emerged that uses mirrors to create a virtual 3D reconstruction system, thereby achieving full surface reconstruction of the sample using a single 3D measurement system. Some scholars have combined a structured light projection system with a reflector to build a 360-degree real-time three-dimensional scanner to obtain complete surface information of the sample. Some scholars have introduced a line laser system into the reflector system, and moved the line laser through a translation stage to achieve three-dimensional scanning, thereby obtaining three-dimensional point cloud data of the sample.

[0004] However, the existing multi-view 3D measurement technology based on plane mirrors still faces some problems:

[0005] 1. Due to the inherent characteristics of DIC and FPP systems, the surface structured light projection system is easily affected by external lighting conditions and the material and texture of the sample surface.

[0006] 2. The laser plane emitted by the flat-scanning line laser is required to be perpendicular to the central axis of the plane mirror. The assembly requirements of the plane mirror and the laser are very high, and it is not suitable for three-dimensional measurement in complex scenes.

[0007] 3. Due to the introduction of the plane mirror, the sample to be tested is imaged in three different areas. The projection light source imaged through the plane mirror will cause light crosstalk at different viewing angles. The fringe pattern captured by the camera contains not only the fringe information used for 3D reconstruction, but also various stray fringes caused by plane mirror reflection.

[0008] 4. Existing multi-view 3D measurement technology is generally only applicable to the 3D reconstruction of small-sized samples. For large-sized objects, it is difficult to achieve low-cost and high-efficiency 3D reconstruction. Summary of the Invention

[0009] Aiming to improve upon existing technologies, the present invention provides a multi-viewpoint line laser scanning device and a method for panoramic measurement of complex surfaces based on binocular stereo vision. By introducing a plane mirror, an object can be imaged from three perspectives, enabling 360-degree panoramic three-dimensional measurement of the surface being measured. A swing-scan laser measurement system based on a rotating stage can simultaneously scan three different areas of a sample's surface, obtaining 3D information about the entire surface in a single scan sequence.

[0010] To achieve the above objectives, the technical methods adopted by the present invention are specifically as follows:

[0011] The present invention discloses a multi-view line laser scanning device based on binocular stereo vision, comprising two cameras, a line laser, a rotating platform, a controller, a computer for data processing connecting the controller and the cameras, and two plane mirrors for collecting multi-view images. The two cameras are at the same height and have a horizontal distance difference. The line laser is fixed on a rotating bracket on the rotating platform. There is a certain vertical distance difference between the line laser and the two cameras. The two plane mirrors for collecting multi-view images are placed behind the object to be measured and form an angle. The rotating platform is connected to the controller.

[0012] As a further improvement, the horizontal distance difference described in the present invention is greater than 10 cm.

[0013] As a further improvement, the vertical distance difference described in the present invention is between 10 cm and 20 cm.

[0014] As a further improvement, the angle between the two plane mirrors described in the present invention is between 90 degrees and 180 degrees.

[0015] As a further improvement, the angle described in the present invention is 120 degrees.

[0016] The present invention discloses a panoramic measurement method of a multi-viewing angle line laser scanning device based on binocular stereo vision, comprising:

[0017] Place the checkerboard calibration plate in N positions, and collect N calibration plate images simultaneously through the primary camera and the secondary camera;

[0018] Extract the coordinates of the corner points in the collected calibration image, and use Zhang's calibration method to calculate the internal and external parameters in the primary and secondary camera coordinate systems, and obtain the pose transformation matrix of the primary and secondary camera coordinate systems and the plane equation of the calibration plate in the primary camera coordinate system;

[0019] Turn on the laser and use the main camera to capture N images of the laser stripes projected onto the calibration plate and the corresponding calibration plate images;

[0020] Use the Steger centerline extraction algorithm to extract the pixel coordinates on the laser stripe image captured by the main camera;

[0021] Use the internal and external parameters in the main camera coordinate system to obtain the plane equation of the calibration plate in the corresponding calibration plate image;

[0022] The 3D coordinates of the laser line are calculated by combining the pixel coordinates of the laser line and the plane equation of the corresponding calibration plate. Finally, the 3D coordinates of the laser line at N different poses are used to fit the laser plane equation in the main camera coordinate system.

[0023] Turn on the turntable. Every time the turntable rotates a certain angle around the rotation axis, turn on the laser and place the calibration plate in N positions. Use the main camera to capture N images of the calibration plate and the laser stripe pattern projected onto the corresponding calibration plate. After obtaining the laser plane equation at M angles, use the laser planes at different angles to fit the rotation axis of the turntable in the main camera coordinate system, and obtain the plane equation of the laser plane at different angles in the main camera coordinate system.

[0024] The calibration plate is placed in N positions. The main camera is used to simultaneously capture the calibration plate and its virtual image in the mirror. The 3D feature point pairs of the calibration plate and the virtual calibration plate image in the mirror are extracted. The plane equation of the plane mirror is then calculated using the Levenberg-Marquardt algorithm with a bundle adjustment strategy to obtain the reflection matrix of the plane mirror in the main camera coordinate system.

[0025] Turn on the rotating stage and laser, and use the primary and secondary cameras to simultaneously capture the laser stripe patterns of the sample in the left plane mirror imaging area, the front surface area, and the right plane mirror imaging area during scanning;

[0026] The laser stripe patterns captured by the primary and secondary cameras are divided into the left plane mirror area, the front surface area, and the right plane area, and the pixel coordinates of the laser lines in each area are extracted;

[0027] Combining the equations of the laser plane moving to different angles in the main camera coordinate system and the pose transformation matrices of the primary and secondary cameras, the feature points corresponding to the primary camera view in the secondary camera view are searched in each area. This allows the laser stripes used for 3D scanning of the sample to be found in the left plane mirror area, the front surface area, and the right plane area in the image captured by the main camera, eliminating stray laser stripes generated by plane mirror reflection.

[0028] Using the obtained main camera intrinsic parameters and the equations of the laser plane moving to different angles in the main camera coordinate system, the laser stripes in the left plane mirror area, the front surface area, and the right plane area used for three-dimensional scanning of the sample in the laser stripe image captured by the main camera are converted into three-dimensional point cloud coordinates;

[0029] The three-dimensional line point cloud coordinates of the virtual surfaces of the left plane mirror area and the right plane mirror area are transformed into the real point cloud coordinates in the main camera coordinate system through the reflection matrix of the plane mirror in the main camera coordinate system;

[0030] The point cloud coordinates obtained in different areas under the main camera coordinate system are merged to obtain the panoramic three-dimensional contour information of the sample surface.

[0031] As a further improvement, the present invention uses laser planes at different angles to fit the rotation axis of the rotating stage in the main camera coordinate system as follows:

[0032] Let the laser plane equation corresponding to the i-th angle in the main camera coordinate system be a i x+b i y+c i z+d i =0;

[0033] The distance from any point on the rotating table axis to the laser plane at different angles is equal. Let G1 = (x1, y1, z1), G2 = (x2, y2, z2), and construct the equation:

[0034]

[0035]

[0036] G1=argmin(g1(G1)+g2(G1)+·······+g N (G1)

[0037] G2=argmin(g1(G2)+g2(G2)+·······+g N (G2)

[0038] Then the direction vector of the rotating table axis is:

[0039]

[0040] The coordinates of any point on the rotating table axis are:

[0041] G1=(x1,y1,z1)

[0042] Minimizing the above formula G={G1G2} is a nonlinear minimization problem, and the parameters of the shaft can be calculated by iteratively solving it using the Levenberg-Marquardt algorithm.

[0043] As a further improvement, the plane equation of the laser plane moving to different angles in the main camera coordinate system of the present invention is specifically obtained as follows:

[0044] Let the plane equation of the laser plane at the initial position of the rotating stage be a0x+b0y+c0z+d0=0, and the plane equation of the laser plane rotating to the i-th position along the axis be a i x+b i y+c i z+d i =0;

[0045] Any point on the light plane at the initial position is denoted as P 0 (x0, y0, z0), the coordinates of the i-th position rotated around the axis are P i (x i ,y i ,z i ), the relationship between the two points can be expressed as:

[0046]

[0047] Among them, T is the rotation transformation matrix between the two coordinate systems, expressed as:

[0048]

[0049] Then when the laser plane rotates to the i-th position, its plane equation parameters can be described as:

[0050] [a i b i c i d i ]=[a0b0c0d0]T(θ) -1

[0051] Among them, θ i is the angle that the laser plane rotates from the initial position to the i-th position.

[0052] As a further improvement, the present invention combines the laser plane equations moving to different angles in the primary camera coordinate system and the pose transformation matrices of the primary and secondary cameras to search for feature points in the secondary camera view corresponding to the primary camera view in each area as follows:

[0053] The rotation transformation matrix of the primary camera coordinate system (X0, Y0, Z0) and the secondary camera coordinate system (X1, Y1, Z1) can be expressed as:

[0054]

[0055] The laser plane equation corresponding to the i-th angle in the main camera coordinate system is Then the laser plane equation parameters corresponding to the i-th angle in the secondary camera coordinate system can be expressed as:

[0056]

[0057] Using the calibrated main camera internal parameters and the light plane equation in the main camera coordinate system, the laser line pixel coordinates extracted from the captured image under the main camera perspective are converted into three-dimensional coordinates in the main camera coordinate system.

[0058] Traverse the pixel coordinates of the laser line extracted from the captured image under the secondary camera's perspective. First, use the calibrated internal parameters of the secondary camera and the light plane equation in the secondary camera coordinate system to convert a certain pixel coordinate into Convert to the 3D coordinates of the secondary camera coordinate system Secondly, the transformation matrix of the primary and secondary cameras is used to transform the three-dimensional point coordinates under the secondary camera's perspective Convert to three-dimensional coordinates in the main camera coordinate system Finally, determine the pixel coordinates under the secondary camera's perspective Whether it matches the pixel coordinates under the main camera's perspective can be determined by:

[0059]

[0060] Where ε is the maximum distance that allows feature points of the primary and secondary camera views to match.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] 1. Simple and convenient installation: The multi-view line laser scanning device proposed in the present invention does not need to meet additional geometric relationships during installation. The light plane emitted by the line laser can be incident at any angle, and the installation requirements are relatively low.

[0063] 2. Simple measurement process: The multi-view line laser scanning measurement method based on binocular stereo vision proposed in the present invention utilizes the principle of binocular matching to solve the problem of light crosstalk between plane mirrors. During the measurement process, there is no need to consider the impact of stray light from the collected image on three-dimensional measurement.

[0064] 3. High measurement efficiency: When the turntable of the multi-view laser scanning device proposed in the present invention rotates, the camera can simultaneously capture laser stripes from three perspectives. Compared with the panoramic measurement based on the turntable or robotic arm, the measurement efficiency is greatly improved.

[0065] 4. The device has low cost and high cost performance. Conventional panoramic three-dimensional measurement often uses a multi-camera device to achieve multi-view measurement. The system is expensive, complex to set up, and tedious to calibrate. The multi-view line laser scanning device proposed in the present invention uses a plane mirror to achieve multi-view imaging. The device has low cost and high cost performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is a structural schematic diagram of the device of the present invention;

[0067] Figure 2 It is a data flow diagram of the method of the present invention;

[0068] Figure 1 In the figure, 1 is the main camera, 2 is the secondary camera, 3 is the line laser, 4 is the high-precision rotation stage, 5 is two plane mirrors, 6 is the object to be measured, 7 is the virtual image of the object to be measured formed by the mirror surface, 8 is the controller, and 9 is the computer for processing data. DETAILED DESCRIPTION

[0069] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0070] Figure 1 Schematic diagram of the device structure of the present invention; the present invention discloses a multi-view line laser scanning device based on binocular stereo vision, characterized in that it includes a main camera 1, a secondary camera 2, a line laser 3, a rotating platform 4, a controller 8, a computer 9 for data processing connecting the controller and the camera, and two plane mirrors 5 for collecting multi-view images. It is required that there is a horizontal distance difference between camera 1 and camera 2 to produce the parallax required for multi-line matching, that the line laser 3 is fixed on a rotating bracket on the rotating platform 4, and that there is a vertical distance difference between the line laser 3 and the two cameras to produce the parallax required for three-dimensional reconstruction. The two mirrors 5 are placed behind the object to be measured and form an angle of approximately 120 degrees. The rotating platform 3 is connected to the controller 8 via a data cable, and the computer 9 is simultaneously connected to the camera and the controller, thereby realizing synchronous control of the camera and the rotating platform.

[0071] The present invention also discloses a panoramic measurement method of a multi-viewing angle line laser scanning device based on binocular stereo vision. Figure 2 It is a data flow diagram of the method of the present invention, comprising:

[0072] Place the checkerboard calibration plate in N positions, and simultaneously capture N calibration plate images using the primary camera 1 and the secondary camera 2.

[0073] Extract the coordinates of the corner points in the collected calibration image, and use Zhang's calibration method to calculate the internal and external parameters in the primary and secondary camera coordinate systems, and obtain the pose transformation matrix of the primary and secondary camera coordinate systems and the plane equation of the calibration plate in the primary camera coordinate system;

[0074] Turn on laser 3 and use the main camera to capture N images of the laser stripe pattern projected by the laser onto the corresponding calibration plate.

[0075] Use the Steger centerline extraction algorithm to extract the pixel coordinates on the laser stripe image captured by the main camera;

[0076] Use the internal and external parameters in the main camera coordinate system to obtain the plane equation of the calibration plate in the corresponding calibration plate image;

[0077] The 3D coordinates of the laser line are calculated by combining the pixel coordinates of the laser line and the plane equation of the corresponding calibration plate. Finally, the 3D coordinates of the laser line at N different poses are used to fit the laser plane equation in the main camera coordinate system.

[0078] Turn on the rotating stage 4. Every time the stage rotates a certain angle around the rotation axis, turn on the laser 3. Place the calibration plate in N positions. Use the main camera to capture N images of the calibration plate and the laser stripe pattern projected onto the corresponding calibration plate. After obtaining the laser plane equation at M angles, use the laser planes at different angles to fit the rotation axis of the rotating stage in the main camera coordinate system.

[0079] Let the plane equation of the laser plane at the initial position of the rotating stage be a0x+b0y+c0z+d0=0, and the plane equation of the laser plane rotating to the i-th position along the axis be a i x+b i y+c i z+d i =0;

[0080] Let the laser plane equation corresponding to the i-th position in the main camera coordinate system be a i x+b i y+c i z+d i =0; the distance from any point on the rotating table axis to the laser plane at different angles is equal. Let G1 = (x1, y1, z1), G2 = (x2, y2, z2), and construct the equation:

[0081]

[0082]

[0083] G1=argmin(g1(G1)+g2(G1)+·······+g N (G1)

[0084] G2=argmin(g1(G2)+g2(G2)+·······+g N (G2)

[0085] Then the direction vector of the rotating table axis is:

[0086]

[0087] The coordinates of any point on the rotating table axis are:

[0088] G1=(x1,y1,z1)

[0089] Minimizing the above formula G={G1G2} is a nonlinear minimization problem, and the parameters of the shaft can be calculated by iteratively solving it using the Levenberg-Marquardt algorithm.

[0090] Use the main camera 1 to simultaneously capture at least 5 real images of the calibration plate in different positions and the virtual image in the mirror, and extract the 3D feature point pairs (real points, virtual points, and virtual images) of the calibration plate and the plane mirror 5 in the image. and virtual points Let G = {a m ,b m ,c m ,d}, we get:

[0091]

[0092] g1(G)=[1-2(a m ) 2 ]x v -2a m b m y v -2a m c m z v +2a m dx r

[0093] g2(G)=-2a m b m x v +[1-2(b m ) 2 ]y v -2b m c m z v +2b m dy r

[0094] g3(G)=-2a m c m x v -2b m c m y v +[1-2(c m ) 2 ]z v +2c m dz r

[0095] Where g1(G), g2(G), g3(G) are residual errors, N is the number of three-dimensional point pairs; limited by the manufacturing quality of the plane mirror 4, the virtual point The accuracy cannot be guaranteed, and a systematic error is introduced into the final calibration result. A bundle adjustment strategy is introduced, and the coordinates of the virtual points are optimized as variables, which is:

[0096]

[0097] Minimizing the above formula is a nonlinear minimization problem, which is solved by the Levenberg-Marquardt algorithm. By solving the above equation, the plane parameters of the plane mirror 5 can be obtained, and then the reflection matrix of the plane mirror 5 in the main camera 1 coordinate system can be obtained. (in

[0098] Turn on the rotating stage 4 and the laser 3, and use the primary and secondary cameras to simultaneously capture the laser stripe patterns of the sample in the left plane mirror imaging area, the front surface area, and the right plane mirror imaging area during scanning;

[0099] The laser stripe patterns captured by the main and auxiliary cameras are divided into the left plane mirror area, the front surface area, and the right plane area, and the pixel coordinates of the laser lines in each area are extracted;

[0100] Combining the laser plane equations moving to different angles in the primary camera coordinate system and the pose transformation matrices of the primary and secondary cameras, search for feature points in the secondary camera view that correspond to the primary camera view in each area.

[0101] The rotation transformation matrix of the primary camera coordinate system (X0, Y0, Z0) and the secondary camera coordinate system (X1, Y1, Z1) can be expressed as:

[0102]

[0103] The laser plane equation corresponding to the i-th angle in the main camera coordinate system is Then the laser plane equation parameters corresponding to the i-th angle in the secondary camera coordinate system can be expressed as:

[0104]

[0105] Using the calibrated main camera internal parameters and the light plane equation in the main camera coordinate system, the laser line pixel coordinates extracted from the captured image under the main camera perspective are converted into three-dimensional coordinates in the main camera coordinate system.

[0106] Traverse the pixel coordinates of the laser line extracted from the captured image under the secondary camera's perspective. First, use the calibrated internal parameters of the secondary camera and the light plane equation in the secondary camera coordinate system to convert a certain pixel coordinate into Convert to the 3D coordinates of the secondary camera coordinate system Secondly, the transformation matrix of the primary and secondary cameras is used to transform the three-dimensional point coordinates under the secondary camera's perspective Convert to three-dimensional coordinates in the main camera coordinate system Finally, determine the pixel coordinates under the secondary camera's perspective Whether it matches the pixel coordinates under the main camera's perspective can be determined by:

[0107]

[0108] Where ε is the maximum distance that allows feature points of the primary and secondary camera views to match.

[0109] Through binocular matching, the laser stripes for 3D scanning of the sample in the left plane mirror area, the front surface area, and the right plane area in the image captured by the main camera are searched, and the stray laser stripes generated by the plane mirror reflection are excluded;

[0110] Using the obtained main camera intrinsic parameters and the equations of the laser plane moving to different angles in the main camera coordinate system, the laser stripes in the left plane mirror area, the front surface area, and the right plane area used for three-dimensional scanning of the sample in the laser stripe image captured by the main camera are converted into three-dimensional point cloud coordinates;

[0111] The three-dimensional line point cloud coordinates of the virtual surfaces of the left plane mirror area and the right plane mirror area are transformed into the real point cloud coordinates in the main camera coordinate system through the reflection matrix of the plane mirror in the main camera coordinate system;

[0112] The point cloud coordinates obtained in different areas under the main camera coordinate system are merged to obtain the panoramic three-dimensional contour information of the sample surface.

[0113] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the core technical features of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A panoramic measurement method based on a multi-view line laser scanning device with binocular stereo vision, characterized in that: This is achieved through the following devices: The system comprises two cameras, a line laser, a rotating stage, a controller, a computer for data processing connected between the controller and the cameras, and two plane mirrors for acquiring multi-view images. The two cameras are at the same height and have a horizontal distance difference. The line laser is fixed to a rotating bracket on the rotating stage. There is a certain vertical distance difference between the line laser and the two cameras. The two plane mirrors for acquiring multi-view images are placed behind the object to be measured and form an angle. The rotating stage is connected to the controller. The horizontal distance difference is greater than 10 cm; the vertical distance difference is between 10 cm and 20 cm; the angle between the two plane mirrors is between 90 degrees and 180 degrees; The panoramic measurement method includes: Place the checkerboard calibration plate in N positions, and collect N calibration plate images simultaneously through the primary camera and the secondary camera; Extract the coordinates of the corner points in the collected calibration image, and use Zhang's calibration method to calculate the internal and external parameters in the primary and secondary camera coordinate systems, and obtain the pose transformation matrix of the primary and secondary camera coordinate systems and the plane equation of the calibration plate in the primary camera coordinate system; Turn on the laser and use the main camera to capture N images of the laser stripes projected onto the calibration plate and the corresponding calibration plate images; Use the Steger centerline extraction algorithm to extract the pixel coordinates on the laser stripe image captured by the main camera; Use the internal and external parameters in the main camera coordinate system to obtain the plane equation of the calibration plate in the corresponding calibration plate image; The 3D coordinates of the laser line are calculated by combining the pixel coordinates of the laser line and the plane equation of the corresponding calibration plate. Finally, the 3D coordinates of the laser line at N different poses are used to fit the laser plane equation in the main camera coordinate system. Turn on the turntable. Every time the turntable rotates a certain angle around the rotation axis, turn on the laser and place the calibration plate in N positions. Use the main camera to capture N images of the calibration plate and the laser stripe pattern projected onto the corresponding calibration plate. After obtaining the laser plane equation at M angles, use the laser planes at different angles to fit the rotation axis of the turntable in the main camera coordinate system, and obtain the plane equation of the laser plane at different angles in the main camera coordinate system. The calibration plate is placed in N positions. The main camera is used to simultaneously capture the calibration plate and its virtual image in the mirror. The 3D feature point pairs of the calibration plate and the virtual calibration plate image in the mirror are extracted. The plane equation of the plane mirror is then calculated using the Levenberg-Marquardt algorithm with a bundle adjustment strategy to obtain the reflection matrix of the plane mirror in the main camera coordinate system. Turn on the rotating stage and laser, and use the primary and secondary cameras to simultaneously capture the laser stripe patterns of the sample in the left plane mirror imaging area, the front surface area, and the right plane mirror imaging area during scanning; The laser stripe patterns captured by the primary and secondary cameras are divided into the left plane mirror area, the front surface area, and the right plane area, and the pixel coordinates of the laser lines in each area are extracted; Combining the equations of the laser plane moving to different angles in the main camera coordinate system and the pose transformation matrices of the primary and secondary cameras, the feature points corresponding to the primary camera view in the secondary camera view are searched in each area. This allows the laser stripes used for 3D scanning of the sample to be found in the left plane mirror area, the front surface area, and the right plane area in the image captured by the main camera, eliminating stray laser stripes generated by plane mirror reflection. Using the obtained main camera intrinsic parameters and the equations of the laser plane moving to different angles in the main camera coordinate system, the laser stripes in the left plane mirror area, the front surface area, and the right plane area used for three-dimensional scanning of the sample in the laser stripe image captured by the main camera are converted into three-dimensional point cloud coordinates; The three-dimensional line point cloud coordinates of the virtual surfaces of the left plane mirror area and the right plane mirror area are transformed into the real point cloud coordinates in the main camera coordinate system through the reflection matrix of the plane mirror in the main camera coordinate system; The point cloud coordinates obtained in different areas under the main camera coordinate system are merged to obtain the panoramic three-dimensional contour information of the sample surface.

2. The panoramic measurement method of the multi-view line laser scanning device based on binocular stereo vision according to claim 1, characterized in that: The specific method of using laser planes at different angles to fit the rotation axis of the rotating stage in the main camera coordinate system is: Let the laser plane equation corresponding to the i-th angle in the main camera coordinate system be a i x+b i y+c i z+d i =0; the distance from any point on the rotating table axis to the laser plane at different angles is equal. Let G1 = (x1, y1, z1), G2 = (x2, y2, z2), and construct the equation: G1=argmin(g1(G1)+g2(G1)+······+g N (G1)) G2=argmin(g1(G2)+g2(G2)+······+g N (G2)) Then the direction vector of the rotating table axis is: The coordinates of any point on the rotating table axis are: G1=(x1,y1,z1) Minimizing the above formula G={G1 G2} is a nonlinear minimization problem. The parameters of the rotating shaft can be calculated by iteratively solving it using the Levenberg-Marquardt algorithm.

3. The panoramic measurement method of the multi-view line laser scanning device based on binocular stereo vision according to claim 1, characterized in that: The plane equations of the laser plane moving to different angles in the main camera coordinate system are specifically obtained as follows: Let the plane equation of the laser plane at the initial position of the rotating stage be a0x+b0y+c0z+d0=0, and the plane equation of the laser plane rotating to the i-th position along the axis be a i x+b i y+c i z+d i =0; Any point on the light plane at the initial position is denoted as P 0 (x0, y0, z0), the coordinates of the i-th position rotated around the axis are P i (x i ,y i ,z i ), the relationship between the two points can be expressed as: Among them, T is the rotation transformation matrix between the two coordinate systems, expressed as: Then when the laser plane rotates to the i-th position, its plane equation parameters can be described as: [a i b i c i d i ]=[a0 b0 c0 d0]T(θ) -1 Among them, θ i is the angle that the laser plane rotates from the initial position to the i-th position.

4. The panoramic measurement method of the multi-view line laser scanning device based on binocular stereo vision according to claim 1, characterized in that: The above-mentioned combination of the laser plane equations moving to different angles in the primary camera coordinate system and the pose transformation matrix of the primary and secondary cameras, searching for the feature points corresponding to the primary camera view in the secondary camera view in each area is specifically as follows: The rotation transformation matrix of the primary camera coordinate system (X0, Y0, Z0) and the secondary camera coordinate system (X1, Y1, Z1) can be expressed as: The laser plane equation corresponding to the i-th angle in the main camera coordinate system is Then the laser plane equation parameters corresponding to the i-th angle in the secondary camera coordinate system can be expressed as: Using the calibrated main camera intrinsic parameters and the light plane equation in the main camera coordinate system, the laser line pixel coordinates extracted from the captured image are converted into three-dimensional coordinates in the main camera coordinate system under the main camera perspective. Traverse the laser line pixel coordinates extracted from the captured image under the secondary camera's perspective. First, use the calibrated secondary camera's internal parameters and the light plane equation in the secondary camera's coordinate system to convert the extracted laser line pixel coordinates into Convert to the 3D coordinates of the secondary camera coordinate system Secondly, the transformation matrix of the primary and secondary cameras is used to transform the three-dimensional point coordinates under the secondary camera's perspective Convert to three-dimensional coordinates in the main camera coordinate system Finally, determine the pixel coordinates under the secondary camera's perspective Whether it matches the pixel coordinates under the main camera's perspective can be determined by: Where ε is the maximum distance that allows feature points of the primary and secondary camera views to match.

Images