A machine vision positioning system and method for robot optical processing

Abstract

The present invention discloses a machine vision positioning system and method for robot optical processing, belonging to the field of optical processing workpiece positioning technology. The positioning system comprises: a robot, a small grinding head processing module, an optical processing platform, an optical element and a machine vision module. The robot is arranged on one side of the optical processing platform, and the small grinding head processing module is installed at the output end of the robot, and is used to process the optical element arranged on the optical processing platform; the basic bracket of the machine vision module is integrated on the side of the optical processing platform, the camera and the light source are integrated on the basic bracket, and the fill light source is arranged on the other side of the optical processing platform. The positioning system uses a light source and a fill light source to eliminate the influence of stray light on the image, obtains the image coordinates of the optical element through image processing captured by the camera, and then obtains the tool coordinates of the optical element through the hand-eye transfer matrix. Rapid and accurate non-contact optical element positioning is achieved, the engineering cycle is reduced, and it has repeatability.

Description

Technical Field

[0001] The present invention relates to the technical field of optical processing workpiece positioning, and in particular to a machine vision positioning system and method for robot optical processing. Background Art

[0002] In recent years, with the maturity of robot technology, more and more six-degree-of-freedom industrial robots are used in the field of large-caliber, medium and low processing precision reflective optical processing, replacing traditional CNC machine tools as motion actuators; compared with CNC machine tools, the advantages of six-degree-of-freedom industrial robots are: low equipment cost, which is conducive to reducing processing costs; high degree of freedom of the equipment, which improves the accuracy of the posture of the processing tool and is conducive to the convergence of surface errors; small footprint and high space utilization, which is conducive to the batch configuration of processing equipment; now industrial robots are mainly combined with optical processing technologies such as small grinding head processing technology and airbag processing technology, which has become the main application of robots in the field of optical processing.

[0003] Machine vision positioning technology has also been gradually improved and is widely used in applications for workpiece detection, such as component defect positioning and feature capture, as well as in applications for tool operation, such as grasping point focusing and path planning, and in multi-task positioning applications such as assisted collaboration. In recent years, with the development of machine learning technology, robots have also been derived to obtain information through machine vision to train models to make autonomous decisions and locate components by themselves.

[0004] Before the robot processes the optical element, the position of the optical element needs to be calibrated, and the conversion relationship between the optical processing platform 3 coordinate system {W} and the robot 1 tool coordinate system {C} is established, so that the robot can clearly know the position of the optical element. At present, the calibration of the position of the optical element mainly adopts the "ruby probe method" of contact repeated measurement. This method requires multiple measurements to be averaged, which affects the accuracy of the position of the optical element during processing. In addition, this method has a long measurement cycle and the risk of potential damage to the component. Summary of the invention

[0005] The purpose of the present invention is to achieve high-precision calibration of optical components in robot processing, combine robot processing with machine vision positioning technology, and provide a machine vision positioning system and method for robot optical processing, which can solve the technical problems existing in the above-mentioned prior art.

[0006] To achieve the above purpose, firstly, the present invention provides a machine vision positioning system for robot optical processing, and the specific technical scheme is as follows:

[0007] The positioning system comprises: a robot, a small grinding head processing module, an optical processing platform and an optical element, wherein the robot is arranged at one side of the optical processing platform, and the small grinding head processing module is installed at the output end of the robot, and is used to perform optical processing on the optical element arranged on the optical processing platform;

[0008] The positioning system also includes a machine vision module, which includes a basic bracket, which is fixed to the side of the optical processing platform and includes: a camera bracket and a light source bracket, the camera bracket is connected to a camera for adjusting the camera shooting area, and the light source bracket is connected to a light source for adjusting the light source irradiation area;

[0009] The positioning system also includes: a fill light source, which is arranged on the side of the optical processing platform and is placed on the opposite side of the light source relative to the optical element.

[0010] Furthermore, the camera bracket includes: a bracket body, the bracket body is vertically connected to the side of the optical processing platform, the bracket body is provided with a lifting mechanism, a locking mechanism is connected below the lifting mechanism, one end of a connecting rod is connected to the lifting mechanism, the other end of the connecting rod is connected to a rotating mechanism, a rotating locking mechanism and a reset mechanism are connected to the rotating mechanism, another connecting rod is connected above the rotating mechanism, a lateral translation mechanism and a lateral locking mechanism are provided below the connecting rod, and a camera is connected below the lateral translation mechanism.

[0011] Furthermore, the light source bracket includes: a square frame and a movable component bracket, the two sides of the square frame are respectively connected to the movable component bracket and the bracket body through movable components, the movable component bracket is vertically connected to the side of the optical processing platform and is arranged parallel to the bracket body, the movable component includes: a pitch mechanism and a pitch locking mechanism, a light emitting plate is provided at the bottom of the square frame, a light source is provided inside the square frame, and a polarizer is covered above the light source.

[0012] In addition, the present invention also provides a machine vision positioning method for robot optical processing, the method comprising the following steps:

[0013] Step 1: Build a machine vision system;

[0014] According to the connection relationship of the machine vision positioning system for robot optical processing described in claims 1-3, a machine vision system is built, and the light source is irradiated to the processing area of ​​the optical processing platform to obtain uniform brightness by adjusting the pitch mechanism, and after completion, the pitch locking mechanism is adjusted to fix the light source; the camera is mounted at the end of the camera bracket, and the rotation mechanism and the lateral translation mechanism of the camera bracket are adjusted to align the camera with the center of the optical processing platform, and then the rotation locking mechanism and the lateral locking mechanism are adjusted to fix the camera, and the fixed point of the reset mechanism is set at this position;

[0015] Step 2: Perform camera calibration;

[0016] Using Zhang Zhengyou calibration method, a calibration plate with a size larger than 1 / 3 of the camera field of view but not larger than the full field of view, a checkerboard size at least three times the length of the camera's working line field of view, and a single side of no less than 5 grids is placed on the optical processing platform to collect at least 20 images from multiple positions and angles; the collected images are imported into the calibration toolbox of Matlab, and the internal and external parameters of the camera coordinate system {C} are output, and the transfer matrix between the camera coordinate system {C} and the optical processing platform coordinate system {W} is obtained;

[0017] Step 3: Perform hand-eye calibration;

[0018] Using the nine-point calibration method, operate the teach pendant to make the robot reach the positions of at least 9 positioning holes on the optical processing platform, and record the position information on the robot teach pendant, then use the calibrated camera to shoot the positioning hole area on the optical processing platform and save the image; perform image processing on the acquired image, use the median filtering method to reduce image noise, use the grayscale enhancement method to improve the image contrast, use the blob algorithm to perform edge detection, and use the geometric centroid method to obtain the image coordinates of the positioning holes in the image and record them; use the nine-point hand-eye calibration method to solve the hand-eye matrix between the robot coordinate system {R} and the camera coordinate system {C} by comparing the above-recorded positioning hole positions in the robot teach pendant with the positioning hole positions in the image obtained by the camera after image processing;

[0019] Step 4: Position the optical components;

[0020] Place the optical components on the optical processing platform, turn off the ambient light, turn on the light, turn on the camera, observe the image quality of the shooting area on the host computer, move the fill light source to eliminate the image, and stop when only the optical components exist in the image in the host computer; shoot images of the optical components, use the grayscale enhancement method to improve the contrast of the acquired images of the optical components, use the blob algorithm to segment the background and obtain the connected area, and obtain a foreground image containing only the outline of the optical components; use the median filtering method, Fourier-Gaussian filtering method, Haar filtering method and other methods to suppress image noise. At this time, since the mirror reflection image of the optical component will appear as an impurity shape in the image area, the edge of the optical component will also be incomplete due to the defects of the blob algorithm. For this, it is necessary to perform an open operation on the component features to isolate the connected areas outside the optical component, and then delete the connected areas that are too small according to the shape threshold, leaving only the incomplete ones. The incomplete outline of the optical element is obtained; at this time, the incomplete outline is locally grown to further complete the original outline, and then a closing operation is performed to make the outline extension become the plane projection of the optical element; then edge extraction is performed using Hough transform, SIFT feature extraction, DOG difference method and other algorithms to obtain the clear outline of the optical element in the image, and Matlab's regionprops method is used to obtain the outline information, and the centroid of the optical element is calculated. The transfer matrix of the camera coordinate system {C} and the optical processing platform coordinate system {W} obtained in step 2 is used to obtain the position of the optical element in the optical processing platform coordinate system {W}. The hand-eye matrix obtained in step 3 is used to calculate the robot coordinate system {R}, and the position of the optical element in the robot coordinate system {R} is obtained; thereby realizing the positioning of the centroid of the projection plane based on robot optical processing machine vision, finding other processing points with the centroid of the projection plane as the feature point, and completing the plane positioning.

[0021] Beneficial effects of the present invention

[0022] Compared with the prior art, the present invention realizes an integrated machine vision module through a machine vision positioning method and utilizes a self-designed bracket, thereby completing the fast, accurate and non-contact positioning of large-aperture optical components; it solves the problems of long positioning time for large-aperture optical components and damage to the mirror surface by the positioning method in optical processing, and realizes fast and damage-free positioning of large-aperture optical components. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A schematic diagram of the overall layout of the machine vision positioning system provided by the present invention;

[0024] Figure 2 Axonometric view of the overall layout of the machine vision positioning system provided by the present invention;

[0025] Figure 3 A schematic diagram of the structure of a camera bracket provided by the present invention;

[0026] Figure 4 A partial enlarged schematic diagram of the camera bracket provided by the present invention;

[0027] Figure 5 A schematic diagram of the structure of a light source support provided by the present invention;

[0028] Figure 6 A schematic diagram of the side structure of the light source bracket provided by the present invention;

[0029] Figure 7 A flow chart of positioning of machine vision optical elements provided by the present invention;

[0030] Figure 8 This is a partial flow chart of the component positioning image processing provided by the present invention.

[0031] In the figure,

[0032] 1. Robot; 2. Small grinding head processing module; 3. Optical processing platform; 4. Optical element; 5. Basic bracket; 6. Camera bracket; 7. Light source bracket; 8. Lifting mechanism; 9. Locking mechanism; 10. Rotating mechanism; 11. Locking mechanism; 12. Reset mechanism; 13. Lateral translation mechanism; 14. Lateral locking mechanism; 15. Camera; 16. Pitch mechanism; 17. Pitch locking mechanism; 18. Light homogenizer; 19. Polarizer; 20. Light source; 21. Calibration plate; 22. Fill light source;

[0033] 61. bracket body; 62. connecting rod;

[0034] 71. Square frame; 72. Movable component bracket. DETAILED DESCRIPTION

[0035] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0036] It should be noted that the terms "top", "bottom", "one side", etc. used in this article to indicate the orientation or position relationship is based on the orientation or position relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description. Similar expressions are only for the purpose of explanation, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention; in addition, the terms "one section" and "two sections" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0037] See also Figure 1-Figure 8 ;

[0038] A machine vision positioning system for robot optical processing, the positioning system comprising: a robot 1, a small grinding head processing module 2, an optical processing platform 3 and an optical element 4, wherein the robot 1 is arranged on one side of the optical processing platform 3, and the small grinding head processing module 2 is installed at the output end of the robot 1, and is used to perform optical processing on the optical element 4 arranged on the optical processing platform 3;

[0039] The positioning system also includes a machine vision module, which includes a basic bracket 5, which is fixed to the side of the optical processing platform 3 and includes: a camera bracket 6 and a light source bracket 7, wherein the camera bracket 6 is connected to a camera 15 for adjusting the shooting area of ​​the camera 15, and the light source bracket 7 is connected to a light source 20 for adjusting the irradiation area of ​​the light source 20; further, the camera bracket 6 includes: a bracket body 61, wherein the bracket body 61 is vertically connected to the side of the optical processing platform 3, a lifting mechanism 8 is provided on the bracket body 61, a locking mechanism 9 is connected below the lifting mechanism 8, one end of a connecting rod 62 is connected to the lifting mechanism 8, and a rotating mechanism 10 is connected to the other end of the connecting rod 62, a rotating locking mechanism 11 and a resetting mechanism 12 are connected to the rotating mechanism 10, another connecting rod 62 is connected above the rotating mechanism 10, a lateral translation mechanism 13 and a lateral locking mechanism 14 are provided below the connecting rod 62, and a camera 15 is connected below the lateral translation mechanism 13.

[0040] Furthermore, the light source bracket 7 includes: a square frame 71 and a movable component bracket 72, the two sides of the square frame 71 are respectively connected to the movable component bracket 72 and the bracket body 61 through movable components, the movable component bracket 72 is vertically connected to the side of the optical processing platform 3 and is arranged parallel to the bracket body 61, the movable component includes: a pitch mechanism 16 and a pitch locking mechanism 17, a light scattering plate 18 is provided at the bottom of the square frame 71, a light source 20 is provided inside the square frame 71, and a polarizer 19 is covered above the light source.

[0041] The positioning system also includes: a fill light source 22, which is arranged on the side of the optical processing platform 3, and is placed on the opposite side of the light source 20 relative to the optical element 4. It should be noted that the light source 20 is a panel light source with a spectrum design to ensure that the mirror surface does not absorb light energy and thus damage the performance of the mirror body, and must match the receiving spectrum of the camera 15 and provide sufficient brightness. The main function is above the optical element 4 to provide wide-angle strong light illumination; the fill light source 22 is an ordinary household vertical fluorescent lamp, whose main purpose is to eliminate the image of the optical element 4 produced by the light source 20 and the ambient stray light, and the main function is below the optical element 4 to provide shadow-eliminating illumination.

[0042] In addition, the present invention also provides a machine vision positioning method for robot optical processing, the positioning method comprising the following steps:

[0043] Step 1: Build a machine vision system;

[0044] A machine vision system is built according to the connection relationship of the machine vision positioning system for robot optical processing as described in claims 1-3, and the light source 20 is adjusted to illuminate the processing area of ​​the optical processing platform 3 to obtain uniform brightness by adjusting the pitch mechanism 16. After completion, the pitch locking mechanism 17 is adjusted to fix the light source 20; the camera 15 is mounted at the end of the camera bracket 6, and the rotation mechanism 10 and the lateral translation mechanism 13 of the camera bracket 6 are adjusted to align the camera 15 with the center of the optical processing platform 3, and then the rotation locking mechanism 11 and the lateral locking mechanism 14 are adjusted to fix the camera 15, and the fixed point of the reset mechanism 12 is set at this position;

[0045] Step 2: Perform camera calibration;

[0046] Using Zhang Zhengyou's calibration method, a calibration plate 21 with a size greater than 1 / 3 of the camera's field of view but not exceeding the full field of view, a checkerboard size of at least three times the length of the camera's 15 working line field of view, and a single side of no less than 5 grids is used, and is placed on the optical processing platform 3 to perform multi-position and multi-angle image acquisition for at least 20 images; the acquired images are imported into the calibration toolbox of Matlab (Matlab→APP→Camera Calibrator), and the internal parameters (focal length and origin displacement) and external parameters (rotation and translation) of the camera 15 coordinate system {C} are output, and the transfer matrix between the camera 15 coordinate system {C} and the optical processing platform 3 coordinate system {W} is obtained;

[0047] Step 3: Perform hand-eye calibration;

[0048] Using the nine-point calibration method, operate the teaching pendant to make the robot 1 reach the positions of at least 9 positioning holes on the optical processing platform 3, and record the position information on the teaching pendant of the robot 1, then use the calibrated camera 15 to shoot the positioning hole area on the optical processing platform 3, and save the image; perform image processing on the acquired image, use the median filtering method to reduce image noise, use the grayscale enhancement method to improve the image contrast, use the blob algorithm to perform edge detection, and use the geometric centroid method to obtain the image coordinates of the positioning holes in the image, and record them; use the nine-point hand-eye calibration method to solve the hand-eye matrix between the robot coordinate system {R} and the camera 15 coordinate system {C} by comparing the above-recorded positioning hole positions in the teaching pendant of the robot 1 with the positioning hole positions in the image obtained by image processing after the camera shooting;

[0049] Step 4: Position the optical components;

[0050] The optical element 4 is placed on the processing platform 3, the ambient light source is turned off, the light source 20 is turned on, the camera 15 is turned on, the image quality of the shooting area is observed on the host computer, the fill light source 22 is moved to eliminate the image, and the process can be stopped when only the optical element 4 exists in the image in the host computer; the optical element 4 is photographed, the grayscale enhancement method is used to enhance the contrast of the image of the acquired optical element 4, the blob algorithm is used to segment the background and obtain the connected area, and a foreground image containing only the outline of the optical element is obtained; the median filtering method, the Fourier-Gaussian filtering method, the Haar filtering method and other methods are used to suppress image noise. At this time, since the mirror reflection image of the optical element 4 will present an impurity shape in the image area, the edge of the optical element 4 will also be incomplete due to the defects of the blob algorithm. For this, it is necessary to perform an open operation on the element features, so that the connected area outside the optical element 4 becomes isolated, and then the too small connected area is deleted according to the shape threshold, leaving only the incomplete The incomplete outline of the optical element 4; at this time, the outline is locally grown for the incomplete part to further complete the original outline, and then a closing operation is performed to make the outline extension become the plane projection of the optical element 4; then edge extraction is performed using Hough transform method, SIFT feature extraction method, DOG difference method and other algorithms to obtain the clear outline of the optical element 4 in the image, and Matlab's regionprops method is used to obtain the outline information, and the centroid of the optical element 4 is calculated. The transfer matrix of the camera 15 coordinate system {C} and the optical processing platform 3 coordinate system {W} obtained in step 2 is used to obtain the position of the optical element 4 in the optical processing platform 3 coordinate system {W}. The hand-eye matrix obtained in step 3 is used to calculate the robot 1 coordinate system {R}, and the position of the optical element 4 in the robot 1 coordinate system {R} is obtained; thereby realizing the positioning of the centroid of the projection plane based on robot optical processing machine vision, finding other processing points with the projection plane centroid as the feature point, and completing the plane positioning.

[0051] After the plane positioning is completed, the optical element processing is carried out. During the processing, the camera 15 is moved away from the camera bracket 6 by loosening the rotation locking mechanism 11, adjusting the rotation mechanism 10, and fixed to a position where it does not interfere with the robot 1. At this point, the positioning task of the optical element 4 is completed. After the optical element processing is completed, check whether the convergence conditions are met. If so, it is delivered; if not, return to step 4 and execute again.

[0052] In addition, after the installation and initial positioning are completed, when the next optical element positioning task begins, it can be implemented from step 4, but first the camera 15 is moved back to the camera bracket 6 by loosening the rotation locking mechanism 11, adjusting the rotation mechanism 10, and using the reset mechanism 12 to reset the camera to the original calibrated position (i.e., the fixed point position of the reset mechanism). The transfer matrix at this position remains unchanged (the camera matrix and hand-eye matrix obtained by steps 2 and 3), so steps 2 and 3 do not need to be repeated.

[0053] The above-mentioned machine vision positioning process of the optical element 4 not only realizes fast, accurate and non-contact positioning of the optical element, but also has positioning repeatability for the optical element 4 by using the light source 20 to eliminate the influence of ambient stray light on imaging, and the fill light source 22 to eliminate the influence of the image caused by the light source 20 irradiating the optical element on imaging. This ensures fast, accurate and non-contact positioning of the optical element 4, reduces the engineering cycle, and avoids contact damage.

Claims

1. A machine vision positioning system for robot optical processing, characterized in that: The positioning system comprises: a robot (1), a small grinding head processing module (2), an optical processing platform (3), and an optical element (4), wherein the robot (1) is arranged on one side of the optical processing platform (3), and the small grinding head processing module (2) is installed at the output end of the robot (1) and is used to perform optical processing on the optical element (4) arranged on the optical processing platform (3); The positioning system further comprises: a basic support (5), the basic support (5) being fixed to the side of the optical processing platform (3), and comprising: a camera support (6) and a light source support (7), the camera support (6) being connected to a camera (15) for adjusting a shooting area of ​​the camera (15), and the light source support (7) being connected to a light source (20) for adjusting an irradiation area of ​​the light source (20); The positioning system further comprises: a fill light source (22), the fill light source (22) being arranged on the side of the optical processing platform (3), and being placed on the opposite side of the light source (20) relative to the optical element (4); The camera bracket (6) comprises: a bracket body (61), the bracket body (61) being vertically connected to the side of the optical processing platform (3), the bracket body (61) being provided with a lifting mechanism (8), the lifting mechanism (8) being connected to a locking mechanism (9) below, the lifting mechanism (8) being connected to one end of a connecting rod (62), the other end of the connecting rod (62) being connected to a rotating mechanism (10), the rotating mechanism (10) being connected to a rotating locking mechanism (11) and a resetting mechanism (12), the rotating mechanism (10) being connected to another connecting rod (62) above, the connecting rod (62) being provided with a lateral translation mechanism (13) and a lateral locking mechanism (14) below, and the lateral translation mechanism (13) being connected to a camera (15) below; The light source bracket (7) comprises: a square frame (71) and a movable component bracket (72); two sides of the square frame (71) are respectively connected to the movable component bracket (72) and the bracket body (61) through movable components; the movable component bracket (72) is vertically connected to the side of the optical processing platform (3) and is arranged parallel to the bracket body (61); the movable component comprises: a pitch mechanism (16) and a pitch locking mechanism (17); a light scattering plate (18) is provided at the bottom of the square frame (71); a light source (20) is provided inside the square frame (71); and a polarizing plate (19) is covered above the light source (20).

2. A machine vision positioning method for robot optical processing, characterized in that: The positioning method comprises the following steps: Step 1: Build a machine vision system; A machine vision system is constructed according to the connection relationship of the machine vision positioning system for robot optical processing as described in claim 1, and the light source (20) is illuminated on the processing area of ​​the optical processing platform (3) by adjusting the pitch mechanism (16) to obtain uniform brightness. After completion, the pitch locking mechanism (17) is adjusted to fix the light source (20); a camera (15) is mounted at the end of the camera bracket (6), and the rotation mechanism (10) and the lateral translation mechanism (13) of the camera bracket (6) are adjusted to align the camera (15) with the center of the optical processing platform (3), and then the rotation locking mechanism (11) and the lateral locking mechanism (14) are adjusted to fix the camera (15), and the fixed point of the reset mechanism (12) is set at this position; Step 2: Perform camera calibration; Using the Zhang Zhengyou calibration method, a calibration plate (21) having a size greater than 1 / 3 of the camera field of view but not greater than the full field of view, a checkerboard size at least three times the length of the working line field of view of the camera (15), and a single side of not less than 5 grids is used, and is placed on the optical processing platform (3) to perform multi-position and multi-angle image acquisition of at least 20 images; the acquired images are imported into the calibration toolbox of Matlab, and the internal parameters and external parameters of the camera (15) coordinate system {C} are output, and thereby the transfer matrix between the camera (15) coordinate system {C} and the optical processing platform 3 coordinate system {W} is obtained; Step 3: Perform hand-eye calibration; Using the nine-point calibration method, operate the teaching pendant to make the robot (1) reach the positions of at least nine positioning holes on the optical processing platform (3), and record the position information on the teaching pendant of the robot (1), then use the calibrated camera (15) to shoot the positioning hole area on the optical processing platform (3), and save the image; perform image processing on the acquired image, use the median filtering method to reduce image noise, use the grayscale enhancement method to improve the image contrast, use the blob algorithm to perform edge detection, use the geometric centroid method to obtain the image coordinates of the positioning holes in the image, and record them; compare the above-recorded positioning hole positions in the teaching pendant of the robot (1) with the positioning hole positions in the image obtained by the camera after image processing, and use the nine-point hand-eye calibration method to solve the hand-eye matrix between the robot coordinate system {R} and the camera (15) coordinate system {C}; Step 4: Position the optical components; The optical element (4) is placed on the optical processing platform (3), the ambient light source is turned off, the light source (20) is turned on, the camera (15) is turned on, the image quality of the shooting area is observed on the host computer, the fill light source (22) is moved to eliminate the image, and the process stops when only the optical element (4) exists in the image on the host computer; the optical element (4) is photographed, the image of the optical element (4) is enhanced by a grayscale enhancement method, the background is segmented by a blob algorithm and a connected area is obtained, and a foreground image containing only the outline of the optical element is obtained; the image noise is suppressed by a median filter method, a Fourier-Gaussian filter method or a Haar filter method. At this time, since the mirror reflection image of the optical element (4) will appear as an impurity shape in the image area, the edge of the optical element (4) will also be incomplete due to the defects of the blob algorithm. For this, an opening operation needs to be performed on the element features so that the connected area outside the optical element (4) becomes isolated, and then the too small connected area is deleted according to the shape threshold, and only the image is removed. The incomplete contour of the incomplete optical element (4) is left behind; at this time, the contour is locally grown at the incomplete part to further complete the original contour, and then a closing operation is performed to make the contour extension become the plane projection of the optical element (4); then edge extraction is performed using the Hough transform method, SIFT feature extraction method or DOG difference method to obtain a clear contour of the optical element (4) in the image, and the regionprops method of Matlab is used to obtain contour information, and the centroid of the optical element (4) is calculated. The transfer matrix of the camera (15) coordinate system {C} and the optical processing platform (3) coordinate system {W} obtained in step 2 is used to obtain the position of the optical element (4) in the optical processing platform (3) coordinate system {W}. The hand-eye matrix obtained in step 3 is used to calculate the robot 1 coordinate system {R}, and the position of the optical element 4 in the robot 1 coordinate system {R} is obtained; thereby realizing the positioning of the centroid of the projection plane based on the robot optical processing machine vision, finding other processing points with the projection plane centroid as the feature point, and completing the plane positioning.

Images