Method and apparatus for analyzing an abrasive material polishing test

Abstract

The application belongs to the field of polishing test, and discloses a grinding material polishing test analysis method and device, the method comprising: S1, placing a test workpiece and a grinding sheet in a polishing test device, and performing multiple rounds of polishing test, each round of polishing test comprising an effective polishing stage and an intermittent control stage; S2, obtaining a surface image of the test workpiece after the polishing test of multiple rounds of cycles is completed; S3, calculating the surface image using an improved omnidirectional multi-scale detection algorithm to obtain a visual performance score of the grinding sheet. The application realizes no-missing response to multiple direction scratches through an omnidirectional filter and continuous direction interpolation, and significantly reduces the missing detection rate of multi-directional cross scratches. Based on the self-adaptive determination of the grinding sheet granularity number, multiple sets of detection scale parameters are determined, so that the filter can effectively capture different width scratches from coarse to fine at the same time, and the comparability of test results of different specifications of materials is ensured.

Description

Technical Field

[0001] This invention relates to the field of grinding testing, and more particularly to a method and apparatus for testing and analyzing grinding materials. Background Technology

[0002] In the field of dry grinding performance testing and analysis of abrasive materials (such as sandpaper, grinding wheels, and other coated abrasives), accurate and quantitative characterization of the scratch morphology formed on the workpiece surface is a key step in evaluating the comprehensive performance of abrasive materials (such as cutting efficiency, wear resistance, and processing consistency). Existing technologies for analyzing surface scratches after grinding mainly rely on traditional image processing and machine vision techniques. However, when applied to standardized grinding testing scenarios, these methods have the following significant shortcomings and limitations.

[0003] In actual grinding processes, due to uneven abrasive grain distribution, equipment vibration, and the superposition of multiple grinding passes, complex scratch networks with multiple directions and varying sizes often form on the workpiece surface. Existing methods mostly use edge detection operators (such as Sobel and Prewitt) or texture filters with a single direction or a limited set of directions, which are difficult to respond to scratch features in any direction simultaneously. This results in a high rate of missed detection of scratches in non-dominant directions or intersecting areas, and the detection results cannot fully reflect the true scratch distribution.

[0004] The scratch width produced by grinding discs of different grit sizes varies significantly. Existing technologies typically use a fixed-scale detection nucleus, which cannot adaptively match the scratch width variations corresponding to coarse-grained (e.g., 80#) to fine-grained (e.g., 1000#) sizes. This results in decreased detection sensitivity for scratches at specific sizes, affecting the comparability of test results between materials of different specifications. Summary of the Invention

[0005] The purpose of this invention is to disclose a method and apparatus for testing and analyzing grinding materials, thereby solving the technical problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: On one hand, the present invention provides a method for testing and analyzing the grinding of abrasive materials, including: S1, Place the test workpiece and grinding disc in the grinding test equipment and perform multiple rounds of grinding tests. Each round of grinding test includes an effective grinding stage and an intermittent control stage. S2, after completing multiple rounds of grinding tests, acquire a surface image of the test workpiece; S3 uses an improved omnidirectional multi-scale detection algorithm to calculate the visual performance score of the grinding disc from the surface image.

[0007] Preferably, the test workpiece and the grinding disc are placed in the grinding test equipment, including: The grinding disc is fixed using the electric grinding head in the grinding test equipment, and the test workpiece is fixed using the workpiece fixture in the grinding test equipment. Before the grinding test begins, the grinding disc is directly above the test workpiece. The electric grinding head can drive the grinding disc to rotate during the effective grinding stage, thereby grinding the test workpiece.

[0008] Preferably, performing multiple rounds of polishing tests includes: Perform a preset number of polishing tests. In each round of polishing tests, if the effective polishing phase ends, the intermittent adjustment phase begins. After the intermittent adjustment phase of the previous round of polishing tests ends, the effective polishing phase of the next round of polishing tests begins. During the intermittent adjustment phase, the electric grinding head stops rotating.

[0009] Preferably, the effective polishing phase duration is the same for each round of polishing test, and the intermittent adjustment phase duration is the same for each round of polishing test.

[0010] Preferably, acquiring a surface image of the test workpiece includes: Take a vertically downward photograph directly above the polished surface of the test workpiece to obtain an image of the workpiece's surface.

[0011] Preferably, an improved omnidirectional multi-scale detection algorithm is used to calculate the visual performance score of the polished disc from the surface image, including: S31, calculate the response value of each pixel in the surface image; S32, based on the response value, obtain the confidence level and dominant direction of each pixel; S33: Calculate the directional response tensor based on the dominant direction, calculate the global response threshold based on the confidence level, and determine whether the pixel is an intersection point based on the directional response tensor and the global response threshold, thereby obtaining the intersection point label value of the pixel. S34, Reconstruct the path of the pixel based on the intersection mark value, dominant direction and confidence to obtain the scratch path set; S35 calculates the visual performance score of the polishing disc based on the scratch path set.

[0012] Preferably, calculating the response value of each pixel in the surface image includes: S30, obtains the directional sampling interval based on the surface image; S31, determine the filter parameters based on the directional sampling interval and the particle size of the grinding disc; S32, generates discrete sampling directions based on the directional sampling interval; S33, based on the filter obtained in S31, calculate the filter response value in the discrete sampling direction; S34, interpolate the filtered response value to obtain the response value of the pixel in the continuous direction.

[0013] Preferably, the directional sampling interval based on the surface image includes: S300 converts the surface image into a grayscale image; S301, calculate the signal-to-noise ratio of the grayscale image; S302, calculates the directional sampling interval based on the signal-to-noise ratio.

[0014] Preferably, the filter parameters are determined based on the directional sampling interval and the grain size of the grinding disc, including: S310, determine the number of sampling directions based on the directional sampling interval; S311, determine the interpolation kernel standard deviation based on the directional sampling interval; S312, Determine the detection scale based on the particle size of the grinding disc; S313 determines the standard deviation, wavelength, and spatial frequency of the filter based on the detection scale.

[0015] On the other hand, the present invention provides a grinding material grinding test and analysis device, including grinding test equipment, imaging equipment and computing equipment; The grinding test equipment is used to perform multiple rounds of grinding tests on the test workpiece and grinding disc. Each round of grinding test includes an effective grinding stage and an intermittent control stage. The imaging equipment is used to acquire surface images of the test workpiece after multiple rounds of grinding tests have been completed; The computing device is used to compute on the surface image using an improved omnidirectional multi-scale detection algorithm to obtain a visual performance score for the grinding disc.

[0016] Beneficial effects: This invention systematically solves the problems mentioned in the background art by introducing an improved omnidirectional multi-scale detection algorithm. Through omnidirectional filtering and continuous direction interpolation, it achieves a complete response to scratches in multiple directions, significantly reducing the false negative rate for multi-directional intersecting scratches. Based on the adaptive determination of multiple sets of detection scale parameters according to the grinding disc grit number, the filter can simultaneously and effectively capture scratches of different widths from coarse to fine, ensuring the comparability of test results for materials of different specifications. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of a grinding material grinding test and analysis method provided by the present invention.

[0019] Figure 2 This is a schematic diagram of the grinding test equipment provided by the present invention.

[0020] Figure 3 This is a schematic diagram of the grinding direction provided by the present invention.

[0021] Figure 4 A comparison chart showing the changes in the weight of the test workpiece and the grinding disc provided by this invention.

[0022] Figure 5 A comparison chart of weight changes provided for this invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0024] like Figure 1 As shown in one embodiment, the present invention provides a method for testing and analyzing the grinding of abrasive materials, comprising: S1, Place the test workpiece and grinding disc in the grinding test equipment and perform multiple rounds of grinding tests. Each round of grinding test includes an effective grinding stage and an intermittent control stage.

[0025] Preferably, the test workpiece and the grinding disc are placed in the grinding test equipment, including: The grinding disc is fixed using the electric grinding head in the grinding test equipment, and the test workpiece is fixed using the workpiece fixture in the grinding test equipment. Before the grinding test begins, the grinding disc is directly above the test workpiece. The electric grinding head can drive the grinding disc to rotate during the effective grinding stage, thereby grinding the test workpiece.

[0026] Specifically, the grinding test equipment of the present invention is as follows: Figure 2 As shown, in Figure 2In the diagram, numbers 1 to 6 represent the grinding disc, test workpiece, electric grinding head, gravity constant pressure device, workpiece fixture, and infrared thermal imager, respectively. The gravity constant pressure device is used to apply pressure to the test workpiece during the grinding test. Figure 2 To facilitate the display of the test workpiece, the test workpiece is drawn above the workpiece fixture. In the actual testing process, the test workpiece is fixed by the workpiece fixture.

[0027] The grinding test equipment of the present invention can be a 300-type planar gravity grinding test equipment.

[0028] Specifically, the grinding disc of the present invention can be a 5-inch, 320# ceramic corundum abrasive grinding disc, with flocked adhesive backing on the reverse side and polyester film sandpaper covering the front side.

[0029] Specifically, the test workpiece of this invention can be an AL6061 aluminum plate, the specifications of which are as follows: It is 150mm long, 100mm wide, and 5mm high, with an initial surface roughness Ra=0.6±0.1μm.

[0030] Specifically, during the grinding process, the gravity constant pressure device applies a pressure of 10N to the test workpiece through the electric grinding head.

[0031] Specifically, during the polishing process, an infrared thermal imager is used to record the temperature of the polished surface of the test workpiece in real time, and an early warning is issued when the temperature exceeds the set temperature threshold.

[0032] After receiving an early warning, staff can pause the grinding process to avoid damage to the grinding and testing equipment due to continuous high temperatures.

[0033] Specifically, such as Figure 3 As shown, during the polishing process, the polishing disc reciprocates along the X and Y axes on the horizontal plane, allowing for polishing of all areas of the surface being polished. Figure 3 The red and green arrows in the middle represent the Y-axis and X-axis directions, respectively.

[0034] Preferably, performing multiple rounds of polishing tests includes: Perform a preset number of polishing tests. In each round of polishing tests, if the effective polishing phase ends, the intermittent adjustment phase begins. After the intermittent adjustment phase of the previous round of polishing tests ends, the effective polishing phase of the next round of polishing tests begins. During the intermittent adjustment phase, the electric grinding head stops rotating.

[0035] Specifically, the preset number of rounds can be set to 4.

[0036] Preferably, the effective polishing phase duration is the same for each round of polishing test, and the intermittent adjustment phase duration is the same for each round of polishing test.

[0037] Specifically, the duration of the effective polishing stage can be set to 5 minutes, and the duration of the intermittent adjustment stage can be set to 5 minutes.

[0038] Preferably, before S1, it further includes: The weights of the grinding disc and the test workpiece were obtained separately.

[0039] Preferably, after each round of grinding test, the dust on the test workpiece and grinding disc is cleaned with a brush or air gun. After the test workpiece and grinding disc have cooled to room temperature, the weight of the test workpiece, sandpaper and grinding disc are measured again.

[0040] like Figure 4 As shown, Figure 4 This is a comparison chart showing the changes in the weight of the test workpiece and the weight of the grinding disc after four rounds of grinding tests on an AL6061 aluminum plate using 5-inch, 320# ceramic corundum abrasive grinding discs.

[0041] S2. After completing multiple rounds of grinding tests, obtain a surface image of the test workpiece.

[0042] Preferably, acquiring a surface image of the test workpiece includes: Take a vertically downward photograph directly above the polished surface of the test workpiece to obtain an image of the workpiece's surface.

[0043] Preferably, after S2, the method further includes obtaining the weight changes of the grinding disc and the test workpiece before and after multiple rounds of grinding tests, and drawing a weight change comparison chart.

[0044] based on Figure 4 By combining the grinding data with the weight change of the tested workpiece, we can obtain the following: Figure 5 The chart showing the weight changes is shown. Figure 5 In the middle, 1T to 4T respectively correspond to Figure 4 From the first 5 minutes to the fourth 5 minutes.

[0045] The effective grinding time T is 5 minutes per session, and the total grinding time is 20 minutes (4T). The amount of workpiece material removed is ΔM_w, and the wear of the grinding disc is ΔM_a. Then: The average grinding efficiency of the grinding disc η = ΣΔM_w / 4T = 1.3605 / 4 = 0.3401g; The average wear of the grinding disc = ΣΔM_a / 4T = 0.0386 / 4 = 0.0096g The wear life of the grinding disc is λ=ΣΔM_w / ΣΔM_a=1.3605 / 0.0386=35.

[0046] S3 uses an improved omnidirectional multi-scale detection algorithm to calculate the visual performance score of the grinding disc from the surface image.

[0047] Preferably, an improved omnidirectional multi-scale detection algorithm is used to calculate the visual performance score of the polished disc from the surface image, including: S31, calculate the response value of each pixel in the surface image; S32, based on the response value, obtain the confidence level and dominant direction of each pixel; S33: Calculate the directional response tensor based on the dominant direction, calculate the global response threshold based on the confidence level, and determine whether the pixel is an intersection point based on the directional response tensor and the global response threshold, thereby obtaining the intersection point label value of the pixel. S34, Reconstruct the path of the pixel based on the intersection mark value, dominant direction and confidence to obtain the scratch path set; S35 calculates the visual performance score of the polishing disc based on the scratch path set.

[0048] This algorithm transforms image pixels into structured scratch paths and quantified scores. Its core benefit lies in achieving a leap from low-level pixel feature detection to high-level morphological structure analysis and comprehensive evaluation. It is no longer limited to statistically calculating the density or width of scratch pixels, but rather reconstructs the macroscopic topological structure of the scratch through a data evolution chain of "dominant direction - confidence level - intersection point - path reconstruction." This allows for the calculation of advanced indicators such as scratch uniformity index and comprehensive performance score, which are directly related to the actual performance of the grinding disc, such as cutting consistency and wear uniformity.

[0049] Preferably, calculating the response value of each pixel in the surface image includes: S30, obtains the directional sampling interval based on the surface image; S31, determine the filter parameters based on the directional sampling interval and the particle size of the grinding disc; S32, generates discrete sampling directions based on the directional sampling interval; S33, based on the filter obtained in S31, calculate the filter response value in the discrete sampling direction; S34, interpolate the filtered response value to obtain the response value of the pixel in the continuous direction.

[0050] This step is the core of the algorithm's omnidirectional and multi-scale detection. Its beneficial effect lies in ensuring the algorithm's high sensitivity to scratches in any direction and strong robustness to different test conditions through parameter adaptation and continuous directional interpolation. Specifically: the directional sampling density is determined adaptively based on image quality, optimizing computational efficiency while ensuring accuracy; the detection scale is preset based on the grinding disc grain size, ensuring that the filter matches the physical size of the scratch; and through discrete sampling followed by interpolation, theoretically, seamless coverage of continuous directions is achieved, solving the problem of missed detections caused by limited directional sampling in traditional methods.

[0051] Preferably, the directional sampling interval based on the surface image includes: S300 converts the surface image into a grayscale image; S301, calculate the signal-to-noise ratio of the grayscale image; S302, calculates the directional sampling interval based on the signal-to-noise ratio.

[0052] This step enables adaptive optimization of the directional sampling interval, improving the algorithm's generalization ability and test stability under different imaging conditions. The directional sampling interval is dynamically adjusted by calculating the image signal-to-noise ratio (SNR): when the image is clear and the SNR is high, a smaller directional sampling interval (more directions) is used to capture subtle directional changes; when the image noise is high, a larger directional sampling interval is used to avoid overfitting noise. This reduces fluctuations in detection performance caused by environmental factors such as lighting and dust, and enhances the repeatability of test results.

[0053] Specifically, the directional sampling interval is calculated based on the signal-to-noise ratio, including: After converting the surface image to a grayscale image, the signal-to-noise ratio of the grayscale image is calculated. Calculate the directional sampling interval based on the signal-to-noise ratio: The minimum directional sampling interval is set to 1°. The maximum directional sampling interval is set to 10°. and Two empirical constants are set to 1 and 0.1 respectively. SNR represents the signal-to-noise ratio of a grayscale image. The directional sampling interval.

[0054] This formula introduces the signal-to-noise ratio as a feedback variable and sets upper and lower limits. and This achieves intelligent and stable control of directional sampling accuracy. The formula ensures that the sampling interval is always within a reasonable range (1° to 10°), preventing the loss of details due to oversampling in high-quality images and avoiding the amplification of noise due to oversampling in low-quality images. The setting of empirical constants balances sensitivity and stability, enabling the algorithm to adapt to most industrial inspection scenarios.

[0055] Preferably, the filter parameters are determined based on the directional sampling interval and the grain size of the grinding disc, including: S310, determine the number of sampling directions based on the directional sampling interval; S311, determine the interpolation kernel standard deviation based on the directional sampling interval; S312, Determine the detection scale based on the particle size of the grinding disc; S313 determines the standard deviation, wavelength, and spatial frequency of the filter based on the detection scale.

[0056] Specifically, the number of sampling directions is determined based on the directional sampling interval, including: The number of sampling directions is calculated using the following formula. : This is the floor function.

[0057] This step is the specific implementation of filter parameter initialization, which establishes a standardized and automated mapping rule from test conditions (image quality, grinding disc specifications) to core algorithm parameters. It ensures three key attributes of subsequent Gabor filter detection: 1) Directional completeness (through...) and 1) Guarantee; 2) Scale matching (guaranteed by granularity query s); 3) Physical interpretability ( , , (All originate from scratch width). This replaces the experience-based parameter tuning in traditional methods, and is key to achieving test standardization and repeatability.

[0058] Specifically, the standard deviation of the interpolation kernel is determined based on the directional sampling interval, including: Calculate the standard deviation of the directional interpolation kernel using the following formula: The standard deviation of the interpolation kernel.

[0059] This formula sets a reasonable smoothing range for directional interpolation, avoiding excessive blurring while maintaining interpolation continuity. Setting the standard deviation of the interpolation kernel to 1.5 times the directional sampling interval is an empirically optimized value. This ensures that the response value of each continuous direction is mainly contributed by the weighted contributions of its nearest discrete direction responses, achieving both directional continuity and maintaining the distinguishability of different directional features.

[0060] Specifically, determining the detection scale based on the particle size of the grinding disc includes: Based on this granularity, the detection scale s is obtained from the preset correspondence table between granularity and detection scale.

[0061] The detection scale is the preset scratch width.

[0062] The benefit of this step is that it directly converts the physical properties (grain size) of the grinding disc into detection parameters (scratch width) that the algorithm can process, thus strongly linking the visual inspection algorithm with the physical process of the grinding test. By querying a preset correspondence table, the algorithm can know how wide the scratch is expected to be on the grinding disc being tested, and adjust the size of the detector accordingly.

[0063] The correspondence table between particle size and detection scale was established based on experimental data. Generally, the smaller the particle size (e.g., 80#), the coarser the abrasive grains, resulting in wider scratches, and the corresponding detection scale should be larger. Conversely, the larger the particle size (e.g., 1000#), the finer the scratches, and the smaller the detection scale. This correspondence allows the algorithm to adaptively match the typical scratch characteristics produced by different grinding discs.

[0064] Specifically, the standard deviation, wavelength, and spatial frequency of the filter are determined based on the detection scale, including: , and These represent the standard deviation, wavelength, and spatial frequency of the filter at the detection scale s, respectively. This represents the width of the scratch.

[0065] Preferably, generating discrete sampling directions based on the directional sampling interval includes: For the k-th discrete sampling direction, the calculation formula is: For the first One discrete sampling direction. This represents the total number of discrete sampling directions.

[0066] Preferably, based on the filter obtained in S31, the filter response value in the discrete sampling direction is calculated, including: For scale and discrete sampling direction The corresponding formula for calculating the filter response value is: This is the filtered response value. For convolution operators, Indicates taking the modulus of a complex number. coordinates The grayscale value of the pixel at that location. The Gabor function is defined as follows: The imaginary unit ( ), The spatial aspect ratio of the Gabor filter is a fixed value. , representing the rotated X-axis coordinate. , represents the rotated Y-axis coordinate, and x and y are the X-axis coordinates and Y-axis coordinates of the pixel, respectively.

[0067] Preferably, interpolation calculation is performed on the filtered response value to obtain the response value of the pixel in continuous directions, including: For continuous directions The response values ​​of a pixel in consecutive directions are obtained by interpolating the discrete sampling response: In scale Continuous direction The response value obtained after interpolation. for The interpolation weights are defined as follows: for and The angular distance between them , This represents the j-th discrete sampling direction. for and The angular distance between them, the calculation method and same.

[0068] The aforementioned interpolation method overcomes the limitations of discrete direction sampling, achieving smooth and seamless response detection across continuous direction space. Using an interpolation method based on Gaussian weights, the algorithm can estimate the response value in any direction. This allows for precise localization of the scratch's true direction (not just a discrete sampling direction) when subsequently searching for the maximum response, significantly improving the accuracy of direction estimation and the ability to detect scratches with arbitrary orientations.

[0069] Preferably, based on the response value, the confidence level and dominant orientation of each pixel are obtained, including: Find the maximum response value and its corresponding parameters: The scale (scale set) that maximizes the response value (one of the elements) The direction (angle value) that maximizes the response value; Indicated in scale ,direction Below, the confidence score of the pixel at coordinates (x, y) will As the dominant direction of this pixel .

[0070] This step assigns two distinct attributes to each pixel: the certainty of the existence of a scratch (confidence) and the most likely direction of the scratch (dominant direction). By searching for the maximum response value across all scale directions, the best-matching scratch "model" for that point is found. The maximum response value itself serves as the confidence C(x,y), quantifying the probability that the point is a scratch; the corresponding direction serves as the dominant direction, representing the dominant orientation of the local texture at that point.

[0071] Preferably, the process involves calculating a directional response tensor based on the dominant direction, calculating a global response threshold based on the confidence level, and determining whether a pixel is an intersection point based on the directional response tensor and the global response threshold, thereby obtaining the intersection point marker value of the pixel, including: For each coordinate Calculate the maximum directional response for each pixel. : This indicates that when the scale is s, in the direction The maximum value of the response value of that pixel obtained from the above.

[0072] Calculate the directional response tensor : For one A real symmetric matrix is ​​used to characterize the energy distribution in local directions. ,express Unit vector on. , where is the directional sampling interval in radians; right Perform eigenvalue decomposition: and For two eigenvalues, satisfying . and There are two feature vectors.

[0073] Eigenvalues and It is through a 2x2 real symmetric matrix It is obtained by eigenvalue decomposition. This is a standard operation in linear algebra.

[0074] For each pixel, its directional response tensor T(x,y) (a 2x2 matrix) has been calculated.

[0075] By finding the roots of the characteristic polynomial of the matrix that are equal to 0, we can obtain the two eigenvalues.

[0076] In practical applications, numerical computation libraries (such as numpy.linalg.eig in NumPy) can be used to directly manipulate matrices. The decomposition process yields eigenvalues ​​and corresponding eigenvectors.

[0077] Calculate the global response threshold : M and W are the number of rows and columns, respectively; The mean of the confidence level. The standard deviation of the confidence level; This is a coefficient used to calculate the global response threshold; for example, it can be set to 0.5. Determine the intersection point: The value is the intersection marker; 1 indicates the coordinate. The pixels are the intersection points, and 0 represents the coordinates. The pixels are non-intersecting points.

[0078] The threshold for intersection detection can be set to, for example, 0.3.

[0079] The above process can accurately identify complex areas where scratches intersect and overlap, distinguishing them from single scratches or noise points. The underlying principle is that at intersections, local texture energy is dispersed in multiple directions, causing the eigenvalues ​​of the directional response tensor T(x,y) to change. and Both are relatively large and the ratios Higher; while in a single scratch area, energy is concentrated in one dominant direction. Very low. Combined with a global response threshold, this step can robustly mark intersections, reducing the probability of edge detection breaking or falsely connecting at intersections.

[0080] Preferably, the path of each pixel is reconstructed based on the intersection marker value, dominant direction, and confidence level to obtain a set of scratch paths: Calculate the global threshold: This represents the maximum confidence level among all confidence levels.

[0081] Calculate the starting threshold for path tracing : Set the starting threshold coefficient for path reconstruction, for example, it can be set to 0.7.

[0082] Calculate the path tracing termination threshold : This is the path reconstruction termination threshold coefficient, which can be, for example, 0.3.

[0083] Select seed point: Select pixels that meet all of the following conditions as path seed points: ; ; exist of Within the neighborhood, The value is the largest.

[0084] Path tracing: Starting from each seed point, trace along its positive direction (the dominant direction of the seed point) and its negative direction (the dominant direction of the seed point + 180°).

[0085] Taking forward tracking as an example: Initialization: Set the current point As the seed point, the dominant direction of the seed point is taken as the current direction. Initialize the list of path points Path=[ ].

[0086] Iterative Prediction and Selection: Step 1, Obtain the prediction point : in, The step size is 1 pixel.

[0087] Step 2, in Within a 3×3 neighborhood, to select candidate points q, the following conditions must be met: The intersection marker value X(q) of candidate point q is 0; The confidence level C(q) of candidate point q is greater than ; ; in, It is the average direction of the vectors of all points in the current Path. The dominant direction of candidate point q; Step 3: If no candidate points are found, terminate the tracking.

[0088] If there are multiple candidate points, calculate the inclusion cost for each candidate point q, and select the candidate point with the smallest inclusion cost as the next prediction point: L represents the total number of direction points in the current Path. α represents the dominant direction of the i-th direction point; α=0.5, β=0.3, γ=0.2 are the weights of the path reconstruction cost function; Add the selected candidate point q to the Path and update. , .

[0089] Termination conditions: encountering a crossover point, confidence level less than or equal to or Greater than or equal to The process will end at that time.

[0090] Step 4: After obtaining the scratch paths for all seed points, merge overlapping scratch paths and remove short paths with a length less than Lmin (e.g., 5 pixels), thus obtaining the scratch path set { }, i = 1, 2, ..., Npath, where Npath is the total number of elements in the scratch path set, and each scratch path is an ordered sequence of pixel coordinates.

[0091] The above scheme can effectively restore the macroscopic geometric shape of the scratch by selecting seed points and using a direction-guided tracking strategy, based on the dominant direction and confidence level, bypassing intersection points, and connecting pixels belonging to the same scratch in an orderly manner to form a scratch path.

[0092] Reverse tracing starts from the same seed point as forward tracing, but its initial direction is set to the dominant direction of that seed point plus 180°. This ensures that tracing proceeds in the opposite direction of the scratch.

[0093] Tracking process: Except for the initial direction, all other steps are exactly the same as forward tracking.

[0094] Path merging: The scratch paths (point sequences) obtained by forward and reverse tracing will be merged into a complete scratch path starting from the seed point.

[0095] Preferably, the visual performance score of the polishing disc is calculated based on the scratch path set, including: Obtain the set of valid pixels: To compare thresholds, This represents the maximum confidence level. This represents the set of valid pixels.

[0096] Will The directional range is evenly divided into A range. This is a preset quantity, for example, it can be 18.

[0097] statistics In the middle, the dominant direction falls within each interval. ( The proportion of pixels The width of interval b is / The first interval is [ / The second interval is [ And so on, the last interval is .

[0098] Calculate directional entropy and direction consistency index : Calculate cross density : ; Divide the grayscale image into For each grid, calculate the number of intersections within each grid. Calculate the mean number of intersections across all grid cells. ; Calculate the standard deviation of the number of intersections within the grid. ; Calculate the coefficient of variation m and n are the number of rows and columns of the grid, respectively.

[0099] Calculate the average path length: ,in For the first The pixel length of the path.

[0100] Calculate the total path length: .

[0101] Calculate the scratch uniformity index : in, =0.6, =0.4, and Two different calculation weights are used in the calculation process of the scratch uniformity index SUI.

[0102] Overall performance score : in =0.5, =0.3 =0.2, representing the three different weights used in the calculation of the overall performance score. In the scoring, the following values ​​are used... Scratch coverage is assessed by its proportion of the image area.

[0103] The above steps, based on the reconstructed path and global statistics, calculate the scratch uniformity index and comprehensive performance score. These indicators quantify the uniformity, directional order, and coverage of scratch distribution, mapping to core performance characteristics of the grinding disc such as cutting stability, wear uniformity, and processing efficiency, providing objective and comprehensive visual dimension data for material evaluation. Specifically, a higher DCI indicates more consistent scratch direction and more ordered texture, corresponding to a more stable and uniform cutting process. Lower cross-density indicates fewer scratch intersections, a clearer and more ordered scratch network, meaning uniform abrasive grain distribution and regular cutting trajectory. A higher ratio of total scratch length to image area indicates more sufficient effective scratch coverage.

[0104] Preferably, in the process of comparing the performance of different grinding discs, they can be sorted from high to low based on the average grinding efficiency, grinding ratio and visual performance scores, resulting in three rankings. Then, the three rankings are weighted to obtain the final ranking.

[0105] The grinding ratio is the result of dividing the amount of workpiece removed by the total wear of the grinding discs. This is a core indicator for measuring lifespan. A higher grinding ratio indicates that the grinding discs are more wear-resistant and have a longer service life.

[0106] By conducting a comprehensive evaluation from different perspectives, it is possible to more accurately compare the performance differences between different grinding discs.

[0107] On the other hand, the present invention provides a grinding material grinding test and analysis device, including grinding test equipment, imaging equipment and computing equipment; The grinding test equipment is used to perform multiple rounds of grinding tests on the test workpiece and grinding disc. Each round of grinding test includes an effective grinding stage and an intermittent control stage. The imaging equipment is used to acquire surface images of the test workpiece after multiple rounds of grinding tests have been completed; The computing device is used to compute on the surface image using an improved omnidirectional multi-scale detection algorithm to obtain a visual performance score for the grinding disc.

[0108] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for testing and analyzing the grinding of abrasive materials, characterized in that, include: S1, Place the test workpiece and grinding disc in the grinding test equipment and perform multiple rounds of grinding tests. Each round of grinding test includes an effective grinding stage and an intermittent control stage. S2, after completing multiple rounds of grinding tests, acquire a surface image of the test workpiece; S3 uses an improved omnidirectional multi-scale detection algorithm to calculate the visual performance score of the grinding disc from the surface image.

2. The method according to claim 1, characterized in that, The test workpiece and grinding disc are placed in the grinding test equipment, including: The grinding disc is fixed using the electric grinding head in the grinding test equipment, and the test workpiece is fixed using the workpiece fixture in the grinding test equipment. Before the grinding test begins, the grinding disc is directly above the test workpiece. The electric grinding head can drive the grinding disc to rotate during the effective grinding stage, thereby grinding the test workpiece.

3. The method according to claim 2, characterized in that, Perform multiple rounds of polishing tests, including: Perform a preset number of polishing tests. In each round of polishing tests, if the effective polishing phase ends, the intermittent adjustment phase begins. After the intermittent adjustment phase of the previous round of polishing tests ends, the effective polishing phase of the next round of polishing tests begins. During the intermittent adjustment phase, the electric grinding head stops rotating.

4. The method according to claim 1, characterized in that, The effective polishing phase duration is the same for each round of polishing tests, and the duration of the intermittent adjustment phase is the same for each round of polishing tests.

5. The method according to claim 1, characterized in that, Obtain surface images of the test workpiece, including: Take a vertically downward photograph directly above the polished surface of the test workpiece to obtain an image of the workpiece's surface.

6. The method according to claim 1, characterized in that, An improved omnidirectional multi-scale detection algorithm is used to calculate the visual performance score of the polished disc from the surface image, including: S31, calculate the response value of each pixel in the surface image; S32, based on the response value, obtain the confidence level and dominant direction of each pixel; S33: Calculate the directional response tensor based on the dominant direction, calculate the global response threshold based on the confidence level, and determine whether the pixel is an intersection point based on the directional response tensor and the global response threshold, thereby obtaining the intersection point label value of the pixel. S34, Reconstruct the path of the pixel based on the intersection mark value, dominant direction and confidence to obtain the scratch path set; S35 calculates the visual performance score of the polishing disc based on the scratch path set.

7. The method according to claim 6, characterized in that, Calculate the response value of each pixel in the surface image, including: S30, obtains the directional sampling interval based on the surface image; S31, determine the filter parameters based on the directional sampling interval and the particle size of the grinding disc; S32, generates discrete sampling directions based on the directional sampling interval; S33, based on the filter obtained in S31, calculate the filter response value in the discrete sampling direction; S34, interpolate the filtered response value to obtain the response value of the pixel in the continuous direction.

8. The method according to claim 6, characterized in that, The directional sampling interval is obtained based on the surface image, including: S300 converts the surface image into a grayscale image; S301, calculate the signal-to-noise ratio of the grayscale image; S302, calculates the directional sampling interval based on the signal-to-noise ratio.

9. The method according to claim 6, characterized in that, The parameters of the filter are determined based on the directional sampling interval and the grain size of the grinding disc, including: S310, determine the number of sampling directions based on the directional sampling interval; S311, determine the interpolation kernel standard deviation based on the directional sampling interval; S312, Determine the detection scale based on the particle size of the grinding disc; S313 determines the standard deviation, wavelength, and spatial frequency of the filter based on the detection scale.

10. A device for testing and analyzing grinding materials, characterized in that, This includes polishing and testing equipment, photography equipment, and computing equipment; The grinding test equipment is used to perform multiple rounds of grinding tests on the test workpiece and grinding disc. Each round of grinding test includes an effective grinding stage and an intermittent control stage. The imaging equipment is used to acquire surface images of the test workpiece after multiple rounds of grinding tests have been completed; The computing device is used to compute on the surface image using an improved omnidirectional multi-scale detection algorithm to obtain a visual performance score for the grinding disc.

Images