A 3D wall thickness extraction method for flexible rubber workpieces based on SAD stereo matching and high-precision equipment collaboration

Abstract

A method for extracting the 3D wall thickness of a flexible rubber workpiece based on SAD stereo matching and collaboration with high-precision equipment. The present invention aims to solve the problem of high-precision measurement of the wall thickness of flexible rubber products, and proposes a method for extracting the 3D wall thickness of a flexible rubber workpiece based on SAD stereo matching and collaboration with high-precision equipment. The method is implemented through the following steps: 1. industrial camera calibration; 2. obtaining the image to be detected; 3. image preprocessing; 4. edge extraction; 5. SAD stereo matching; 6. collaborative splicing with high-precision equipment based on image features; 7. 3D determination of feature wall thickness; 8. testing the robustness of the measurement method and numerical consistency analysis. The present invention combines high-precision equipment control and image processing algorithms in the field of visual measurement of flexible materials, achieving high-precision shooting and splicing detection of various types of workpieces, overcoming the difficulties of large errors in conventional manual measurement and poor robustness of general measurement methods for various types of workpieces, with high accuracy and high speed.

Description

Technical Field

[0001] The present invention relates to a method for extracting the 3D wall thickness of a flexible rubber workpiece based on SAD stereo matching and collaboration with high-precision equipment. Background Art

[0002] With the continuous development of society and the economy, competition in the automotive market is becoming increasingly fierce. Modern auto parts production involves a large number of components, and the quality of each component indirectly determines the vehicle's reputation for quality. Plastic dust boots, a crucial component of automotive parts, have their quality impacting production orders, making quality inspection a crucial process in their production. However, rubber dust boots, cast from a flexible, bellows-shaped material, are difficult to automate due to their unique shape and material. Traditionally, measurement involves cutting the workpiece along its generatrix after casting and manually measuring the cut surface with a vernier caliper. This process is tedious and time-consuming, and manual measurement can easily lead to inaccurate measurement points and an unequal number of measurements. Consequently, there is a growing demand for automated measurement of plastic dust boots.

[0003] Computer vision technology has developed rapidly internationally. Since the mid-1960s, research on computer vision has intensified, particularly in the field of binocular stereo vision. Various novel methods for stereo vision research have been proposed and applied. Robert of MIT was the first to extend previous two-dimensional image research to three-dimensional scenes, using computers to analyze and study images in three dimensions. This marked a pioneering effort in stereo vision research. Since then, an increasing number of scholars have devoted themselves to the study of stereo vision, which has, to a certain extent, promoted the development of stereo vision technology. Although domestic research in the computer vision industry, especially binocular vision technology, started relatively late compared to other countries, many researchers have devoted themselves to these fields in recent years and achieved relatively advanced results.

[0004] In recent years, binocular vision systems have been increasingly used in measurement applications. However, their application to measuring small, flexible targets at close range is relatively limited. Therefore, this paper specifically studies the use of binocular stereo vision technology to measure the wall thickness of flexible rubber workpieces. Summary of the Invention

[0005] The present invention combines high-precision equipment control and image processing algorithms in the field of visual measurement of flexible materials, overcoming the difficulties of large errors in conventional manual measurement and poor robustness of general measurement methods for various types of workpieces. It has high accuracy and fast speed, greatly improving measurement accuracy and efficiency, and proposes a 3D wall thickness extraction method for flexible rubber workpieces based on SAD stereo matching and high-precision equipment collaboration.

[0006] The above-mentioned object of the invention is achieved through the following technical solutions:

[0007] Step 1: Use a dot calibration plate to perform multiple image calibration and distortion correction, and then perform epipolar correction to obtain the internal and external parameters and relative pose of the binocular camera;

[0008] Step 2: Automatically calculate the high-precision slide trajectory based on the workpiece length, and use the binocular camera to capture the workpiece in sections to obtain images;

[0009] Step 3: Preprocess the image and obtain ROI;

[0010] Step 4: Use edge extraction algorithm to extract the edge of the obtained ROI;

[0011] Step 5: Use the SAD algorithm to find the corresponding points of the left and right camera features;

[0012] Step 6: Perform stitching based on image features and verify by combining motion trajectory analysis of a high-precision slide;

[0013] Step 7: Use the edge profile to calculate the 3D wall thickness of the feature position;

[0014] Step 8: Verify the robustness of the measurement method and whether the measurement results meet the consistency requirements.

[0015] Effects of the invention:

[0016] The present invention utilizes SAD stereo matching and feature registration for binocular vision inspection. The main challenges of the present invention include the collaborative splicing of high-precision equipment and the adaptability to the measurement of a variety of flexible rubber workpieces. It has the following advantages:

[0017] 1. This invention does not rely on precise image coordinates: The detection method uses the workpiece's inherent characteristics as a benchmark, using edge extraction and SAD stereo matching features. Based on these image features, it collaborates with high-precision equipment for stitching and verification. This ensures algorithm stability. Even if the mechanical structure of the detection system has certain errors or the captured image has some offset, it can still stably stitch images and calculate the workpiece's 3D thickness.

[0018] 2. High Accuracy: Because non-standard flexible rubber components are prone to deformation, a traversal algorithm is used to calculate the 3D wall thickness of feature points, obtaining extreme values ​​and deriving the actual wall thickness. This method also combines SAD stereo matching with high-precision equipment collaborative splicing to ensure detection accuracy, as shown in Table 1.

[0019] 3. Robust Algorithm: By detecting the inherent features of flexible rubber workpieces, automatically locating the joint positions, and combining this with high-precision slide control for calibration, this algorithm automatically stitches images of different workpiece models and calculates 3D wall thickness at all feature locations. This algorithm completely eliminates the impact of mechanical installation and runtime errors. This flexibility ensures easy updating and maintenance of the visual inspection system. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a flow chart of a method for extracting 3D wall thickness of a flexible rubber workpiece based on SAD stereo matching and high-precision equipment collaboration;

[0021] Figure 2 This is a schematic diagram of the device structure proposed in the first embodiment, in which 1, high-precision slide; 2, servo motor; 3, bracket; 4, cylinder; 5, fixture; 6, binocular camera system;

[0022] Figure 3 It is a local image of the workpiece taken by the camera in different areas as proposed in the first embodiment;

[0023] Figure 4 It is a local image of the workpiece taken by the camera in different areas as proposed in the first embodiment;

[0024] Figure 5 It is a local image of the workpiece taken by the camera in different areas as proposed in the first embodiment;

[0025] Figure 6 This is a partial ROI obtained by using image preprocessing proposed in the first embodiment;

[0026] Figure 7 This is the XLD graph extracted using the Canny operator proposed in the first embodiment;

[0027] Figure 8 It is a complete workpiece image spliced ​​based on image features proposed in the first embodiment;

[0028] Figure 9 It is a local image of the workpiece taken by the camera in different areas according to the specific embodiment 1;

[0029] Figure 10 It is a local image of the workpiece taken by the camera in different areas according to the specific embodiment 1;

[0030] Figure 11 It is a local image of the workpiece taken by the camera in different areas according to the specific embodiment 1;

[0031] Figure 12 It is a local image of the workpiece taken by the camera in different areas according to the specific embodiment 1;

[0032] Figure 13 This is a partial ROI obtained by using image preprocessing proposed in the first embodiment;

[0033] Figure 14 This is the XLD graph extracted using the Canny operator proposed in the first embodiment;

[0034] Figure 15 It is a complete workpiece image spliced ​​based on image features proposed in the first embodiment; DETAILED DESCRIPTION

[0035] Specific embodiment 1: This embodiment is a method for extracting the 3D wall thickness of a flexible rubber workpiece based on SAD stereo matching and high-precision equipment collaboration, which is specifically prepared according to the following steps:

[0036] Step 1: Use a dot calibration plate to perform multiple image calibration and distortion correction to obtain the internal and external parameters and relative pose of the binocular camera;

[0037] Step 2: According to the length of the workpiece, the high-precision slide trajectory is automatically calculated, and the binocular camera shoots the workpiece in sections to obtain images such as Figure 3 、 Figure 4 、 Figure 5 ;

[0038] Step 3: Use pre-processing methods such as binarization, smoothing filtering, image enhancement, opening and closing operations to obtain ROI, such as Figure 6 ;

[0039] Step 4: Use the Canny algorithm to extract the edge of the ROI. Figure 7 ;

[0040] Step 5: Use the SAD algorithm to find the corresponding points of the left and right camera features;

[0041] Step 6: stitching based on image features and combining with high-precision slide motion trajectory analysis for verification, such as Figure 8 ;

[0042] Step 7: Use the edge profile to calculate the 3D wall thickness of the feature position;

[0043] Step 8: Verify the robustness of the measurement method and whether the measurement results meet the consistency requirements;

[0044] Effects of this implementation:

[0045] This implementation method utilizes SAD stereo matching and high-precision equipment to perform binocular vision inspection. The main difficulties of this implementation method include the coordinated splicing of high-precision equipment and the adaptability to the measurement of various flexible rubber workpieces. This implementation method also has the following advantages:

[0046] 1. This implementation method does not rely on precise image coordinates: The detection method uses the workpiece's inherent features as a benchmark, using edge extraction and SAD stereo matching features. Based on these image features, it collaborates with high-precision equipment for stitching and verification. This ensures algorithm stability. Even if the mechanical structure of the inspection system has certain errors and the captured images exhibit some offset, the images can still be stably stitched and the workpiece's 3D thickness calculated.

[0047] 2. High Accuracy: Because non-standard flexible rubber components are prone to deformation, a traversal algorithm is used to calculate the 3D wall thickness at feature points, obtaining extreme values ​​and deriving the actual wall thickness. This method also combines SAD stereo matching with image feature registration and splicing to ensure detection accuracy by calculating the 3D wall thickness at the feature locations, as shown in Table 1.

[0048] 3. Robust Algorithm: By detecting the characteristic positions of flexible rubber workpieces, automatically locating the splicing locations, and combining PLC-controlled high-precision slide calibration, this algorithm automatically stitches images of different workpiece models and calculates the 3D wall thickness at all characteristic locations. This algorithm completely eliminates the impact of mechanical installation and runtime errors. The flexibility of this method ensures easy updating and maintenance of the visual inspection system.

[0049] Specific embodiment 2: This embodiment differs from specific embodiment 1 in that, in step 2, the high-precision slide's trajectory is automatically calculated based on the workpiece length, and the binocular camera captures the workpiece in sections to acquire images. Specifically, the high-pixel binocular camera has a small repeating field of view. To cope with flexible rubber workpieces with complex lengths and features, the automatic high-precision slide's trajectory is controlled and externally triggered by the binocular camera. The flexible rubber workpiece is segmented and photographed to acquire the images required for stitching. Simultaneously, the length of the high-precision slide's trajectory is recorded, providing verifiable data for subsequent stitching based on image features. Other steps and parameters are the same as those in specific embodiment 1.

[0050] Specific embodiment 3: This embodiment differs from specific embodiments 1 and 2 in that: in step 5, the SAD algorithm is used to find the corresponding feature points of the left and right cameras. The absolute values ​​of the corresponding pixel differences of the corresponding pixel blocks of the aligned left and right view images are summed:

[0051]

[0052] Among them, d∈[d min , dmax ], where d min and d max The maximum and minimum values ​​of the same point in the space of the two images can be obtained by specifying the length of the disparity search area as 1=d max -d min After calculating the similarity for a point in the entire disparity search area, the disparity at the minimum or maximum metric value is used as the matching point for that point. The other steps and parameters are the same as those in the first and second embodiments.

[0053] Specific embodiment 4: This embodiment differs from specific embodiments 1 to 3 in that: in step 6, stitching is performed based on image features, and motion trajectory analysis of the high-precision slide is used to perform verification. The specific process is as follows:

[0054] (1) Perform extreme value extraction on the XLD extracted by the Canny algorithm to extract the position information of the feature points in each group of contours;

[0055] (2) Based on the extracted feature point coordinate information and the relationship between the PLC's precise control of the high-precision slide motion distance, the coordinate information of the same feature point in the two images is matched. Using the extracted feature point information, a spatial transformation algorithm is used to remove the redundant parts, and the same feature points in the two images are combined to complete the overall image stitching.

[0056] (3) The PLC precisely controls the movement distance of the high-precision slide, converts the distance into a pixel value in the pixel coordinate system, and compares it with the intercepted redundant part. If the value is the same, the splicing is successful. The other steps and parameters are the same as those in the first to third embodiments.

[0057] Specific embodiment 5: This embodiment differs from specific embodiments 1 to 4 in that in step 7, the edge contour is used to calculate the 3D wall thickness at the feature position. Since the flexible rubber workpiece is prone to deformation, a traversal method is used to calculate the 3D wall thickness of the feature point. The specific process is as follows:

[0058] (1) The coordinate information of the feature points at the same position in the binocular camera is combined with the relative pose of the participants in the binocular camera to obtain the actual space X, Y, and Z coordinates of any point on the map.

[0059] (2) Taking the feature point on the right edge of the flexible rubber workpiece as the reference, its coordinates are (x1, y1, z1), find the same point as x1 on the left edge, and extract the xyz coordinates of N points above and below, a total of 2N+1 point coordinates.

[0060] (3)Use Traverse the above 2N+1 points and put their coordinates into the formula (x n ,y n, z n ), and get Thick min The other steps and parameters are the same as those in the first to fourth embodiments.

[0061] Specific embodiment 6: This embodiment differs from specific embodiments 1 to 5 in that: in step 8, the robustness of the measurement method and whether the measurement results meet the consistency requirements are tested. The specific steps are as follows:

[0062] (1) Follow the steps to check whether different types of flexible rubber workpieces can be completely spliced ​​together;

[0063] (2) The same workpiece is tested 8 to 10 times continuously to check the range of the group values ​​of the same feature point. If the consistency reaches ±0.03, the method is effective. The other steps and parameters are the same as those in the specific embodiments 1 to 5. The following examples are used to verify the beneficial effects of the present invention:

[0064] Example 1:

[0065] This embodiment provides a method for extracting the 3D wall thickness of a flexible rubber workpiece based on SAD stereo matching and high-precision equipment collaboration, which is specifically prepared according to the following steps:

[0066] Step 1: Use a dot calibration plate to perform multiple image calibration and distortion correction, and then perform epipolar correction to obtain the internal and external parameters and relative pose of the binocular camera:

[0067] (1) Use the binocular camera to shoot the calibration plate to obtain the left and right eye images, and use the calibration program to complete the calibration of the binocular camera to obtain the camera's intrinsic parameters and the relative position relationship between the two cameras;

[0068] (2) According to the calibration relationship between the ideal projection point and the distorted projection point of the standard template image, the distortion coefficient of the camera distortion model can be obtained. Assume that (x, y) is the actual position of the distortion point on the imaging plane. According to

[0069]

[0070] k1, k2, and k3 represent the first, second, and third order of radial distortion, respectively. P1 and P2 are the projections of a target point in space onto the two camera coordinate systems. The camera coordinates of the image are corrected using the distortion coefficients. After correction, the camera coordinate system is converted to the image pixel coordinate system using the intrinsic parameter matrix, and the new image coordinates are assigned according to the pixel values ​​of the source image coordinates.

[0071] (3) The pixel coordinate system of the image with distortion correction is converted into the camera coordinate system through the intrinsic parameter matrix (compared with the physical coordinate system of the image, it has more scaling and Z axis), and the parallel epipolar correction is performed through the rotation matrices R1 and R2. Then, the camera coordinates of the image are corrected by the distortion coefficient. After correction, the camera coordinate system is converted into the image pixel coordinate system through the intrinsic parameter matrix, and the new image coordinates are assigned according to the pixel values ​​of the source image coordinates;

[0072] Step 2: According to the length of the workpiece, the high-precision slide trajectory is automatically calculated, and the binocular camera shoots the workpiece in sections to obtain images, such as Figure 9 , Figure 10 , Figure 11 , Figure 12 Specifically, the high-pixel binocular camera has a small repetitive field of view. To cope with flexible rubber workpieces with complex lengths and features, a PLC is used to control a high-precision slide and a binocular camera. Based on the length and features of the workpiece, the flexible rubber workpiece is automatically divided into sections and photographed to obtain the images required for stitching. At the same time, the trajectory of the high-precision slide under PLC control between each shot is recorded to provide verifiable data for the subsequent stitching based on image feature registration.

[0073] Step 3: Use pre-processing methods such as binarization, smoothing filtering, image enhancement, opening and closing operations to obtain ROI, such as Figure 13 ;

[0074] Step 4: Use the Canny algorithm to extract the edge of the ROI. Figure 14 :

[0075] (1) Use Gaussian smoothing filter to suppress the noise that obeys the normal distribution, and use I σ =I*G σ , where * represents the convolution operation; G σ is a two-dimensional Gaussian kernel with standard deviation σ, defined as:

[0076] (2) Use the Soble horizontal and vertical operators to convolve with the input image to calculate dx and dy:

[0077]

[0078]

[0079] Then, d x =f(x, y)*Sobel x (x, y)d y =f(x, y)*Sobel y (x, y). The magnitude of the image gradient can be further obtained: M(x, y) = |d x (x, y)|+|dy (x, y)|, angle is: θ M =arctan(d y / d x ).

[0080] (3) Perform NMS processing based on the gradient direction obtained in the previous step. Compare the image pixel with its neighboring pixels along the gradient direction or in the opposite direction. If it is the maximum value, it is retained; otherwise, it is suppressed, that is, the pixel is set to 0. That is:

[0081] N[i,j]=NMS(M[i,j],ζ[i,j])

[0082] (4) Set the high and low thresholds. The two thresholds can filter out noise in the image and improve the image quality. The principle of weak edge processing is that the weak edge points of the real edge all have strong edge points;

[0083] Step 5: Use the SAD algorithm to find the corresponding points of the left and right camera features. Sum the absolute values ​​of the corresponding pixel differences of the corresponding pixel blocks of the aligned left and right view images:

[0084]

[0085] Among them, d∈[d min , d max ], where d min and d max The maximum and minimum values ​​of the same point in the space of the two images can be obtained by specifying the length of the disparity search area as 1=d max -d min +1. After calculating the similarity for a point in the entire disparity search area, the disparity at the minimum or maximum metric value is used as the matching point for that point;

[0086] Step 6: stitching based on image features and combining with high-precision slide motion trajectory analysis for verification, such as Figure 15 :

[0087] (1) Perform extreme value extraction on the contour lines extracted by the Canny algorithm to extract the position information of the feature points in each group of contours;

[0088] (2) Based on the extracted feature point coordinate information and the relationship between the PLC's precise control of the high-precision slide motion distance, the coordinate information of the same feature point in the two images is matched. Using the extracted feature point information, a spatial transformation algorithm is used to remove the redundant parts, and the same feature points in the two images are combined to complete the overall image stitching.

[0089] (3) According to the PLC, the high-precision slide moves a distance that is accurately controlled, and the distance is converted into a pixel value in the pixel coordinate system. The distance is compared with the intercepted redundant part. If the values ​​are the same, the splicing is successful.

[0090] Step 7: Use the edge contour to calculate the 3D wall thickness of the feature position. Since the flexible rubber workpiece is prone to deformation, a traversal method is used to calculate the 3D wall thickness of the feature point. The specific process is as follows:

[0091] (1) The coordinate information of the feature points at the same position in the binocular camera is combined with the relative pose of the participants in the binocular camera to obtain the actual space X, Y, and Z coordinates of any point on the map.

[0092] (2) Taking the feature point on the right edge of the flexible rubber workpiece as the reference, its coordinates are (x1, y1, z1), find the same point as x1 on the left edge, and extract the xyz coordinates of 50 points above and below, a total of 101 point coordinates.

[0093] (3)Use Traverse the above 101 points and put their coordinates into the formula (x n ,y n , z n ), and get Thick min This is the 3D wall thickness of the feature point.

[0094] Step 8: Verify the robustness of the measurement method and whether the measurement results meet the consistency requirements:

[0095] (1) Follow the steps to check whether different types of flexible rubber workpieces can be completely spliced ​​together;

[0096] (2) The same workpiece is tested 8 to 10 times continuously to check whether the range of the group values ​​of the same feature point is consistent to ±0.03 (range 0.06). This method is effective, as shown in Table 1. The feature points are selected according to the actual needs of the user as shown in the following table:

[0097]

[0098] Table 1 Example of product test data containing 5 peaks and 5 troughs

[0099] SAD stereo matching is a global matching based on pixel grayscale value and calibration parameter correction, which effectively matches all feature corresponding points in the image; the splicing method based on the collaboration of features and high-precision equipment is a high-efficiency and high-accuracy method based on the actual needs of customers and the characteristics of the rubber dustproof workpiece itself that are measured. By using these two points, flexible rubber workpieces with different lengths and different numbers of feature points can be effectively spliced ​​together, which has strong robustness; PLC controls high-precision equipment to coordinate splicing measurements and perform splicing verification, which is one of the innovations of this invention; the wall thickness is obtained in three dimensions in space, the measurement consistency is judged, the accuracy of the measurement method is verified, and the problem of 3D wall thickness extraction of flexible rubber workpieces is successfully solved.

Claims

1. A method for extracting 3D wall thickness of flexible rubber workpieces based on SAD stereo matching and high-precision equipment collaboration, characterized by: The method is specifically carried out according to the following steps: Step 1: Use a dot calibration plate to perform multiple image calibration and distortion correction, and then perform epipolar correction to obtain the internal and external parameters and relative pose of the binocular camera; Step 2: Automatically calculate the high-precision slide trajectory based on the workpiece length, and use the binocular camera to capture the workpiece in sections to obtain images; Step 3: Preprocess the image and obtain ROI; Step 4: Use edge extraction algorithm to extract the edge of the obtained ROI; Step 5: Use the SAD algorithm to find the corresponding points of the left and right camera features; Step 6: Perform stitching based on image features and verify by combining motion trajectory analysis of a high-precision slide; Step 7: Use the edge profile to calculate the 3D wall thickness of the feature position; Step 8: Verify the robustness of the measurement method and whether the measurement results meet the consistency requirements.

2. The method for extracting 3D wall thickness of a flexible rubber workpiece based on SAD stereo matching and high-precision equipment collaboration according to claim 1, characterized in that: In step 2, the high-precision slide trajectory is automatically calculated based on the length of the workpiece, and the binocular camera shoots the workpiece in segments to obtain images. Specifically, the high-pixel binocular camera has a small repeated field of view. In order to cope with flexible rubber workpieces with complex lengths and features, the automatic high-precision slide trajectory and the binocular camera are controlled by external triggering, and the flexible rubber workpiece is divided into sections and photographed to obtain the images required for splicing. At the same time, the length of the trajectory traveled by the high-precision slide is recorded to provide verifiable data for the subsequent splicing based on image features.

3. The method for extracting 3D wall thickness of a flexible rubber workpiece based on SAD stereo matching and high-precision equipment collaboration according to claim 1, characterized in that: In step 5, the SAD algorithm is used to find the corresponding feature points of the left and right cameras, and the absolute values ​​of the corresponding pixel differences of the corresponding pixel blocks of the aligned left and right view images are summed: Among them, d∈[d min , d max ], where d min and d max The maximum and minimum values ​​of the same point in the space of the two images can be obtained by specifying the length of the disparity search area as l = d max -d min +1, after calculating the similarity for a point in the entire disparity search area, use the disparity at the minimum or maximum metric value as the matching point for that point.

4. The method for extracting 3D wall thickness of a flexible rubber workpiece based on SAD stereo matching and high-precision equipment collaboration according to claim 1, characterized in that: In step 6, stitching is performed based on image features, and verification is performed using motion trajectory analysis of the high-precision slide. The specific process is as follows: (1) Perform extreme value extraction on the XLD extracted by the edge detection algorithm to extract the position information of the feature points in each group of contours; (2) According to the extracted feature point coordinate information and the relationship between the precise control of the high-precision slide motion distance, the coordinate information of the same feature point in the two images is matched. Through the extracted feature point information, the redundant part is cut off using the spatial transformation algorithm, and the same feature points in the two images are combined to complete the overall image stitching work; (3) According to the precise control of the high-precision slide moving distance, the distance is converted into a pixel value in the pixel coordinate system and compared with the intercepted redundant part. If the value is the same, the splicing is successful.

5. The method for extracting 3D wall thickness of a flexible rubber workpiece based on SAD stereo matching and high-precision equipment collaboration according to claim 1, characterized in that: In step 7, the edge contour is used to calculate the 3D wall thickness of the feature position. Since the flexible rubber workpiece is prone to deformation, a traversal method is used to calculate the 3D wall thickness of the feature point. The specific process is as follows: (1) The coordinate information of the feature points at the same position in the binocular camera is combined with the relative pose of the participants in the binocular camera to obtain the actual space x, y, and z coordinates of any point on the map; (2) Taking the characteristic point on the right edge of the flexible rubber workpiece as the reference, its coordinates are (x1, y1, z1), find the same point as x1 on the left edge, and extract the x, y, and z coordinates of n points above and below, for a total of 2n+1 point coordinates; (3)Use Traverse the above 2n+1 points and put their coordinates into the formula (x n ,y n , z n ), and get Thick min This is the 3D wall thickness of the feature point.

6. The method for extracting 3D wall thickness of a flexible rubber workpiece based on SAD stereo matching and high-precision equipment collaboration according to claim 1, characterized in that: In step eight, the robustness of the measurement method and whether the measurement results meet the consistency requirements are tested. The specific steps are as follows: (1) Follow the steps to check whether different types of flexible rubber workpieces can be completely spliced ​​together; (2) The same workpiece is tested 8 to 10 times continuously to check the range of the group values ​​of the same feature point. If the consistency reaches ±0.03, the method is effective.

Images