Method for measuring passivation radius of cutting edge of indexable blade based on binocular vision

Abstract

The invention discloses an indexable blade cutting edge passivation radius measuring method based on binocular vision, and belongs to the technical field of binocular vision measurement. The method comprises the following steps: calibrating left and right cameras by using a binocular vision system to obtain internal and external parameters and distortion coefficients of the cameras; secondly, acquiring an indexable blade cutting edge passivation image for image correction, so that the corrected images are in the same plane and are parallel to each other; secondly, improving the blade image quality through a wavelet packet transformation image preprocessing method, and reducing the noise influencing the image quality; thirdly, obtaining an image disparity map through a region-based optimization SGBM stereo matching algorithm, and changing the disparity map into a depth map according to a triangulation principle; and finally, measuring the passivation radius of the cutting edge of the blade on the basis of the obtained depth map, and comparing the passivation radius of the cutting edge of the blade with the passivation radius of the cutting edge of the blade measured by an Alcona three-dimensional detection instrument so as to judge the precision of the research.

Description

Technical Field

[0001] The invention relates to the field of binocular vision measurement, and in particular to a binocular vision-based method for measuring the blunting radius of an indexable insert cutting edge. Background Art

[0002] In the field of tool manufacturing and cutting processing, the use of indexable inserts (such as carbide inserts, ceramic inserts, etc.) is extremely common. The edge passivation layer, as an oxide layer formed on the tool surface under complex working conditions such as friction, thermal effects, or chemical reactions, has a significant impact on tool performance. The width of the passivation layer directly affects the tool's cutting performance, tool life, and stability during the cutting process. The appropriate passivation layer width can optimize the tool's cutting force distribution, reduce tool wear, and improve the quality of the machined surface; conversely, an inappropriate passivation layer width will lead to a decrease in tool cutting performance, accelerated tool wear, and even cause processing errors, affecting product quality.

[0003] Traditional methods for measuring the passivation layer on tool edges include physical contact measurement and optical non-contact measurement. While physical contact measurement methods, such as using a microscope or contact probe to measure the width of the cutting edge, offer high accuracy, they are prone to tool damage and measurement instability during the contact measurement process. Furthermore, the measurement process is susceptible to interference from external factors, resulting in poor measurement stability and difficulty meeting the requirements for high-precision and high-reliability measurements.

[0004] Optical measurement methods, such as laser scanning or digital image processing, avoid the physical damage associated with contact measurement, but they still face challenges such as limited measurement accuracy, surface reflection issues, and difficulty processing complex surface shapes. Laser scanning technology has certain limitations in measurement accuracy, making it difficult to meet the growing demand for high-precision measurement. Furthermore, the reflective properties of the tool surface can interfere with the reception and processing of the laser signal, leading to increased measurement errors. Digital image processing technology, when processing tools with complex surface shapes, faces difficulties in image feature extraction and limited measurement accuracy, making it impossible to accurately obtain precise information about the edge passivation layer.

[0005] Therefore, binocular vision-based measurement methods have emerged. Binocular vision technology can obtain depth information of an object by calculating the difference between two images from different perspectives, thereby achieving accurate three-dimensional measurement. The application of binocular vision in measuring the width of the blade edge passivation layer is expected to overcome the shortcomings of traditional methods, providing high-precision, non-contact, and real-time measurement results, meeting the stringent requirements of modern manufacturing for tool quality inspection. Summary of the Invention

[0006] The present invention aims to provide a binocular vision-based method for measuring the blunting radius of an indexable insert cutting edge, so as to solve the technical problems of low working efficiency and insufficient precision in the prior art.

[0007] In order to solve the above technical problems, the present invention specifically provides the following technical solutions:

[0008] A binocular vision-based method for measuring the blunting radius of an indexable insert cutting edge comprises the following:

[0009] Step S1: Build a binocular vision system and set parameters for the experiment;

[0010] Step S2: calibrate the binocular cameras using a checkerboard grid based on the established binocular vision system to obtain the intrinsic and extrinsic parameters and distortion coefficients of the two cameras;

[0011] Step S3, capturing left and right images of the indexable insert's edge passivation using a binocular vision system, and performing stereoscopic correction of the images using the data from step S2 to ensure that the epipolar lines of the left and right cameras remain horizontal in the corrected images;

[0012] Step S4, pre-processing the captured blade passivation image to reduce noise and enhance image contrast;

[0013] Step S5, generating a disparity map of the image obtained in step S4 based on the optimized SGBM stereo matching algorithm in the region, and obtaining a depth map of the blade passivation image according to the triangulation principle and the formula Z=(f*b) / d;

[0014] Step S6: Based on the depth map obtained in step S5, the blunt radius of the blade edge is measured and compared with the blunt radius of the blade edge measured by the Alicona 3D detection instrument to determine the accuracy of the research method.

[0015] The step S2 comprises the following steps:

[0016] Step S2-1, placing the chessboard required for the experiment under the left and right camera lenses;

[0017] Step S2-2, shooting at least 20 sets of chessboard images at different angles. Angle change can be achieved by moving the chessboard;

[0018] Step S2-3, start calibration using the Stereo Camera Calibration toolbox in MATLAB;

[0019] Step S2-4: Import the image from step S2-2 and set the checkerboard size. MATLAB automatically detects the corner points of the checkerboard and deletes one or more groups of images with large errors. Repeat step S2-3 until the error meets the calibration requirements.

[0020] Step S2-5, derive parameters, which include the internal parameters f of the two cameras x ,f y ,c x ,c y , the external parameter rotation matrix R, the translation vector T, the distortion coefficients are radial distortion K1, K2, K3, and tangential distortion P1, P2.

[0021] The left and right images of the indexable insert edge passivation layer are captured by a binocular vision system, and the data from step S2 are used to perform stereo correction on the image. Stereo correction is also called epipolar correction. Its purpose is to obtain the mapping matrix required for stereo correction, and then perform distortion correction and stereo correction on the image. The correction method is the Bouguet epipolar correction method.

[0022] The image stereo correction method in step S3 is as follows:

[0023] Corrected image = original image * matrix M;

[0024] Matrix M = Matrix N * Matrix R * Matrix N -1

[0025] The matrix N is composed of the camera intrinsic parameters obtained through camera calibration, and its expression is

[0026]

[0027] where f x ,f y is the focal length of the left and right cameras in the x and y directions, c x ,c y is the position of the camera center;

[0028] The matrix R is the camera's external parameter, which includes the rotation matrix R and the translation vector T. Its expression is:

[0029]

[0030] Matrix N -1 is the inverse matrix of matrix N;

[0031] Any pixel P after image correction rect It can be mapped from the original image pixel point P:

[0032] P rect =M*P.

[0033] The step S4 comprises the following steps:

[0034] Step S4-1, converting the stereo-rectified image in step S3 into a grayscale image;

[0035] Step S4-2: Using different denoising methods such as mean filtering, median filtering, Gaussian filtering, and wavelet packet denoising, comparing the processed results, and selecting the wavelet packet filtering method to reduce the noise in the image;

[0036] Step S4-3, performing edge detection at the sub-pixel level of the image based on the Zernike moment. The nth-order and mth-order Zernike moment of the two-dimensional function f(xy) is defined as:

[0037]

[0038] Where * represents conjugation, V nm is a Zernike polynomial, defined as:

[0039] V nm (xy)=V nm (ρθ)=R nm (ρ)exp(jmθ)x 2 +y 2 ≦1

[0040] Where m and n are non-negative integers n-|m| is an even number, |m|≦n, R nm (ρ) is the radial polynomial;

[0041] When nm is an odd number

[0042] When nm is an even number, R nm (ρ)=0.

[0043] The region-based optimized SGBM stereo matching algorithm comprises the following steps:

[0044] Step S5-1, pre-processing: Use the horizontal Sobel operator to perform edge detection on the left and right blade passivation layer images to obtain the gradient image. The horizontal Sobel operator formula is as follows:

[0045] Sobel(xy)=2*[p(x+1y)-p(x-1y)]+p(x+1y-1)-p(x-1y-1)+p(x+1y+1)-p(x-1y+1);

[0046] In the formula, p is pixel;

[0047] Step S5-2, matching cost calculation: For each pixel, calculate its matching cost with the corresponding pixel at different disparities using the sum of absolute differences (SAD) as the cost function. The function is as follows: C SAD (pd)=Σ q∈Np ∣I L(p)-I R (pd)∣;

[0048] In the formula, the pixel to be matched in the left camera image is p, the disparity is d, and the corresponding pixel in the right camera image is pd. Step S5-3, energy function minimization: For each pixel, an energy function is defined, including a data term and a smoothing term. The data term represents the matching cost, and the smoothing term represents the disparity continuity of adjacent pixels. The dynamic programming method is to calculate the cumulative cost along multiple directions and find the minimum value as the final cost. The disparity selection of each pixel is determined by a winner-takes-all dynamic programming formula as follows:

[0049]

[0050] In the formula, L is the cumulative cost function of the current path, P1 and P2 are two parameters that control the parallax smoothness, and P1 must be less than P2;

[0051] Step S5-4, disparity map generation: for each pixel, the optimal disparity is selected according to the final cost and a disparity map is generated;

[0052] Step S5-5, disparity optimization: repair or fill abnormal values or holes in the disparity map using uniqueness detection and sub-pixel interpolation methods, and optimize the disparity map obtained in step S5-4 using WLS filtering.

[0053] The depth map of the blade passivation image is obtained according to the triangulation principle and the formula Z=(f*b) / d. The specific derivation process is as follows:

[0054] For a certain pixel point P at the center C of the left camera l and right camera center C l The imaging points on the upper image plane are P l and P r , the left and right cameras are in the same plane with the left boundary of the left plane as the starting point P l and P r The distance d between the points and the corresponding camera plane is l and d r , define the disparity d as d l -d r ;

[0055] Remember C l C r The distance between them is b, P l P r The distance is m, the depth is Z, the focal length is f, and the parallax is d. According to the principle of similar triangles, we have:

[0056]

[0057] Solving the simultaneous equations yields Z = (f*b) / d.

[0058] The beneficial effects of the present invention are:

[0059] Using binocular vision technology to measure the blunt radius of indexable inserts, researchers are developing a three-dimensional image of the insert's blunt edge. This allows for more accurate measurement of the blunt radius, enabling timely determination of tool replacement and minimizing machining errors caused by tool damage. This non-contact measurement method improves the efficiency of insert quality inspection while preserving the insert's integrity. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 Schematic diagram of the flow of the method for measuring the width of the passivation layer of the cutting edge of an indexable insert based on binocular vision according to an embodiment of the present invention;

[0061] Figure 2 Schematic diagram of image stereo correction in an embodiment of the present invention;

[0062] Figure 3 Flowchart of the region-based optimized SGBM stereo matching algorithm in an embodiment of the present invention;

[0063] Figure 4 This is a real picture of Alicona used for comparison in the embodiments of the present invention. DETAILED DESCRIPTION

[0064] The following is a detailed description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of this application and not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative work are within the scope of protection of this invention.

[0065] Example 1

[0066] This embodiment provides a method for measuring the blunting radius of an indexable insert edge based on binocular vision. Figure 1 This is a flow chart of a method for measuring the blunt radius of an indexable insert edge according to an embodiment of the present invention. Figure 1 This embodiment is described;

[0067] Step S1: Build a binocular vision system and set the corresponding parameters for the experiment;

[0068] In this embodiment, the binocular vision system is composed of left and right cameras, lenses, apertures, and power cables;

[0069] Step S2: calibrate the binocular cameras using a checkerboard grid based on the built binocular vision system to obtain the intrinsic and extrinsic parameters and distortion coefficients of the two cameras;

[0070] Specifically, step S2 includes the following steps:

[0071] Step S2-1, placing the chessboard required for the experiment under the left and right camera lenses;

[0072] Step S2-2, shooting at least 20 sets of chessboard images at different angles. Angle change can be achieved by moving the chessboard;

[0073] Step S2-3, start calibration using the Stereo Camera Calibration toolbox in MATLAB;

[0074] Step S2-4: Import the image from step S2-2 and set the checkerboard size. MATLAB automatically detects the corner points of the checkerboard and deletes one or more groups of images with large errors. Repeat step S2-3 until the error meets the calibration requirements.

[0075] Step S2-5, derive parameters, which include the internal parameters f of the two cameras x ,f y ,c x ,c y , external parameter rotation matrix R, translation vector T, distortion coefficients include radial distortion K1, K2, K3, tangential distortion P1, P2;

[0076] Step S3, capturing left and right images of the indexable insert's edge passivation using a binocular vision system, and performing stereoscopic correction of the images using the data from step S2 to ensure that the epipolar lines of the left and right cameras remain horizontal in the corrected images;

[0077] Figure 2 This is a schematic diagram of image stereo correction in an embodiment of the present invention, ensuring that the epipolar lines of the left and right cameras remain horizontal in the corrected image;

[0078] Taking left and right images of the indexable insert edge passivation by a binocular vision system, and performing stereoscopic correction of the images using the data from step S2;

[0079] Stereo correction is also called epipolar correction. Its purpose is to obtain the mapping matrix required for stereo correction, and then perform distortion correction and stereo correction on the image. The most common correction method is the Bouguet epipolar correction method.

[0080] The principle of image stereo correction is as follows:

[0081] Corrected image = original image * matrix M, matrix M = matrix N * matrix R * matrix N -1 ;

[0082] The matrix N is composed of the camera intrinsic parameters obtained through camera calibration and its expression is:

[0083]

[0084] where f, x f y is the focal length of the left and right cameras in the x and y directions, c x ,c y is the position of the camera center;

[0085] The matrix R is the camera external parameter, with the rotation matrix R and translation vector T, and its expression is

[0086]

[0087] Matrix N -1 is the inverse matrix of matrix N;

[0088] Any pixel P after image correction rect It can be mapped from the original image pixel point P:

[0089] P rect =M*P;

[0090] Step S4, pre-processing the captured blade passivation image to reduce noise and enhance image contrast;

[0091] Specifically, step S4 includes the following steps:

[0092] Step S4-1, converting the stereo-rectified image in step S3 into a grayscale image;

[0093] Step S4-2: Using different denoising methods such as mean filtering, median filtering, Gaussian filtering, and wavelet packet denoising, comparing the processed results, and selecting the wavelet packet filtering method to reduce the noise in the image;

[0094] Step S4-3, performing edge detection at the sub-pixel level of the image based on the Zernike moment. The nth-order and mth-order Zernike moment of the two-dimensional function f(xy) is defined as:

[0095]

[0096] Where * represents conjugation, V nm is a Zernike polynomial, defined as:

[0097] V nm (xy)=V nm (ρθ)=R nm (ρ)exp(jmθ)x 2 +y 2 ≦1;

[0098] Where m and n are non-negative integers n-|m| is an even number, |m|≦n, Rnm (ρ) is a radial polynomial.

[0099]

[0100] When nm is an even number, R nm (ρ) = 0;

[0101] Step S5, generating a disparity map of the image obtained in step S4 based on the optimized SGBM stereo matching algorithm in the region, and obtaining a depth map of the blade passivation image according to the triangulation principle and the formula Z=(f*b) / d;

[0102] Step S6, measuring the blunt radius of the blade edge based on the depth map obtained in step S5, and comparing it with the blunt radius of the blade edge measured by the Alicona 3D detection instrument to determine the accuracy of the research method;

[0103] The SGBM stereo matching algorithm process is as follows Figure 3 As shown, it specifically includes the following steps:

[0104] Step S5-1, pre-processing: Use the horizontal Sobel operator to perform edge detection on the left and right blade passivation layer images to obtain the gradient image. The horizontal Sobel operator formula is as follows:

[0105] Sobel(xy)=2*[p(x+1y)-p(x-1y)]+p(x+1y-1)-p(x-1y-1)+p(x+1y+1)-p(x-1y+1);

[0106] In the formula, p is pixel.

[0107] Step S5-2, matching cost calculation: For each pixel, the matching cost with the corresponding pixel at different disparities is calculated using the sum of absolute differences (SAD) as the cost function as follows:

[0108] C SAD (pd)=∑ q∈Np ∣I L (p)-I R (pd)∣;

[0109] In the formula, the pixel to be matched in the left camera image is p, the disparity is d, and the corresponding pixel in the right camera image is pd.

[0110] Step S5-3, energy function minimization: For each pixel, an energy function is defined, including a data term and a smoothing term. The data term represents the matching cost, and the smoothing term represents the disparity continuity of adjacent pixels. The dynamic programming method is to calculate the cumulative cost along multiple directions and find the minimum value as the final cost. The disparity selection of each pixel is determined by a winner-takes-all dynamic programming formula as follows:

[0111]

[0112] In the formula, L is the cumulative cost function of the current path, P1 and P2 are two parameters that control the parallax smoothness, and P1 must be less than P2;

[0113] Step S5-4, disparity map generation: for each pixel, the optimal disparity is selected according to the final cost and a disparity map is generated;

[0114] Step S5-5, disparity optimization: repair or fill abnormal values or holes in the disparity map using uniqueness detection and sub-pixel interpolation methods, and optimize the disparity map obtained in step S5-4 using WLS filtering;

[0115] Figure 4 Schematic diagram of the triangulation principle in an embodiment of the present invention. According to the following derivation process, the formula Z = (f*b) / d can be obtained;

[0116] The right camera is in the same plane and takes the left boundary of the left plane as the starting point, P l and P r The distance d between the points and the corresponding camera plane is l and d r , define the disparity d as d l -d r ;

[0117] Remember C l C r The distance between them is b, P l P r The distance is m, the depth is Z, the focal length is f, and the parallax is d. According to the principle of similar triangles, we have:

[0118]

[0119] Solving the simultaneous equations yields Z = (f*b) / d.

[0120] It should be emphasized that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Various modifications and improvements may be made to the present invention without departing from the novel spirit and scope of the present invention, and such modifications and improvements are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for measuring the blunting radius of an indexable insert edge based on binocular vision, characterized in that: Including the following; Step S1: Build a binocular vision system and set parameters for the experiment; Step S2: calibrate the binocular cameras using a checkerboard grid based on the established binocular vision system to obtain the intrinsic and extrinsic parameters and distortion coefficients of the two cameras; Step S3, capturing left and right images of the indexable insert's edge passivation using a binocular vision system, and performing stereoscopic correction of the images using the data from step S2 to ensure that the epipolar lines of the left and right cameras remain horizontal in the corrected images; Step S4, pre-processing the captured blade passivation image to reduce noise and enhance image contrast; Step S5, generating a disparity map of the image obtained in step S4 based on the optimized SGBM stereo matching algorithm in the region, and obtaining a depth map of the blade passivation image according to the triangulation principle and the formula Z=(f*b) / d; Step S6: Based on the depth map obtained in step S5, the blunt radius of the blade edge is measured and compared with the blunt radius of the blade edge measured by the Alicona 3D detection instrument to determine the accuracy of the research method.

2. The method for measuring the blunting radius of an indexable insert cutting edge based on binocular vision according to claim 1, characterized in that: The step S2 comprises the following steps: Step S2-1, placing the chessboard required for the experiment under the left and right camera lenses; Step S2-2, shooting at least 20 sets of chessboard images at different angles. Angle change can be achieved by moving the chessboard; Step S2-3, start calibration using the Stereo Camera Calibration toolbox in MATLAB; Step S2-4: Import the image from step S2-2 and set the checkerboard size. MATLAB automatically detects the corner points of the checkerboard and deletes one or more groups of images with large errors. Repeat step S2-3 until the error meets the calibration requirements. Step S2-5, derive parameters, which include the internal parameters f of the two cameras x ,f y ,c x ,c y , the external parameter rotation matrix R, the translation vector T, the distortion coefficients are radial distortion K1, K2, K3, and tangential distortion P1, P2.

3. The method for measuring the blunting radius of an indexable insert cutting edge based on binocular vision according to claim 1, characterized in that: The left and right images of the indexable insert edge passivation layer are captured by a binocular vision system, and the data from step S2 are used to perform stereo correction on the image. Stereo correction is also called epipolar correction. Its purpose is to obtain the mapping matrix required for stereo correction, and then perform distortion correction and stereo correction on the image. The correction method is the Bouguet epipolar correction method.

4. The method for measuring the width of the passivation layer of the indexable insert cutting edge based on binocular vision according to claim 1, characterized in that: The image stereo correction method in step S3 is as follows: Corrected image = original image * matrix M; Matrix M = Matrix N * Matrix R * Matrix N -1 The matrix N is composed of the camera intrinsic parameters obtained through camera calibration, and its expression is where f x ,f y is the focal length of the left and right cameras in the x and y directions, c x ,c y is the position of the camera center; The matrix R is the camera's external parameter, which includes the rotation matrix R and the translation vector T. Its expression is: Matrix N -1 is the inverse matrix of matrix N; Any pixel P after image correction rect It can be mapped from the original image pixel point P: P rect =M*P。 5. The binocular vision-based indexable insert edge blunting radius method according to claim 1, characterized in that: The step S4 comprises the following steps: Step S4-1, converting the stereo-rectified image in step S3 into a grayscale image; In step S4-2, different denoising methods such as mean filtering, median filtering, Gaussian filtering, and wavelet packet denoising are used. After comparing the processed results, the wavelet packet filtering method is selected to reduce the noise in the image. In step S4-3, edge detection at the sub-pixel level of the image is performed based on the Zernike moment. The nth-order mth-order Zernike moment of the two-dimensional function f(xy) is defined as: Where * indicates conjugation, V nm is a Zernike polynomial, defined as: V nm (xy)=V nm (ρθ)=R nm (ρ)exp(jmθ)x 2 +y 2 ≦1 Where m and n are non-negative integers and n-|m| is an even number. R nm (ρ) is the radial polynomial; When nm is an odd number When nm is an even number, R nm (ρ)=0.

6. The method for measuring the blunting radius of an indexable insert cutting edge based on binocular vision according to claim 1, characterized in that: The region-based optimized SGBM stereo matching algorithm comprises the following steps: Step S5-1, pre-processing: Use the horizontal Sobel operator to perform edge detection on the left and right blade passivation layer images to obtain the gradient image. The horizontal Sobel operator formula is as follows: Sobel(xy)=2*[p(x+1y)-p(x-1y)]+p(x+1y-1)-p(x-1y-1)+p(x+1y+1)-p(x-1y+1); In the formula, p is pixel; Step S5-2, matching cost calculation: For each pixel, the matching cost with the corresponding pixel at different disparities is calculated using the sum of absolute differences (SAD) as the cost function as follows: C SAD (pd)=∑ q∈Np ∣I L (p)-I R (p-d)∣; In the formula, the pixel to be matched in the left camera image is p, the disparity is d, and the corresponding pixel in the right camera image is pd. Step S5-3, energy function minimization: For each pixel, an energy function is defined, including a data term and a smoothing term. The data term represents the matching cost, and the smoothing term represents the disparity continuity of adjacent pixels. The dynamic programming method is to calculate the cumulative cost along multiple directions and find the minimum value as the final cost. The disparity selection of each pixel is determined by a winner-takes-all dynamic programming formula as follows: In the formula, L is the cumulative cost function of the current path, P1 and P2 are two parameters that control the parallax smoothness, and P1 must be less than P2; Step S5-4, disparity map generation: for each pixel, the optimal disparity is selected according to the final cost and a disparity map is generated; Step S5-5, disparity optimization: repair or fill abnormal values or holes in the disparity map using uniqueness detection and sub-pixel interpolation methods, and optimize the disparity map obtained in step S5-4 using WLS filtering.

7. The method for measuring the blunting radius of an indexable insert cutting edge based on binocular vision according to claim 1, characterized in that: The depth map of the blade passivation image is obtained according to the triangulation principle and the formula Z=(f*b) / d. The specific derivation process is as follows: For a certain pixel point P at the center C of the left camera l and right camera center C r The imaging points on the upper image plane are P l and P r , the left and right cameras are in the same plane with the left boundary of the left plane as the starting point P l and P r The distance d between the points and the corresponding camera plane is l and d r , define the disparity d as d l -d r ; Note C l C r The distance between them is b, P l P r The distance is m, the depth is Z, the focal length is f, and the parallax is d. According to the principle of similar triangles, we have: Solving the simultaneous equations yields Z = (f*b) / d.

Images