Diamond saw blade cutting edge precision measurement system based on laser scanning

Abstract

The present application relates to the field of visual measurement, in particular to a diamond saw blade cutting edge precision measurement system based on laser scanning, the system comprises: speckle feature extraction module, used for controlling single-wavelength laser beam oblique illumination of saw blade cutting edge end face, calling industrial camera to collect corresponding light and dark spot image pixel matrix of matrix surface, and calculating to obtain local speckle intensity contrast index.In the present application, the fractal dimension index of topography is calculated and compared with the preset recession threshold, which can push the static topography result to the wear failure judgment level, so that the change of chip space, the change of cutting edge complexity and the change of recession trend can obtain unified measurement basis, which not only enhances the correlation ability of measurement result to the real use state, but also enhances the direct decision value of result output, helps to reduce one-sided judgment generated by single image observation, and improves the comprehensive evaluation ability of cutting edge precision state, wear degree and failure time.

Description

Technical Field

[0001] This invention relates to the field of visual measurement technology, and in particular to a laser scanning-based precision measurement system for diamond saw blade cutting edges. Background Technology

[0002] Visual measurement technology is a non-contact measurement technology system based on optical imaging, image processing, and spatial geometric analysis. It mainly acquires two-dimensional or three-dimensional image information of the object being measured through optical sensing devices such as cameras and lasers, and then performs analytical calculations on the pixel distribution, grayscale changes, edge features, and spatial structure in the image to achieve accurate measurement of size, shape, position, and surface characteristics.

[0003] The shortcomings of existing technologies lie in the lack of a continuous transmission relationship from surface scattering response to roughness distribution and then to spatial morphology degradation. This leads to a disconnect between image features and the actual cutting conditions during practical operation. For example, under the same contour size, the cutting edge face may already show micro-wear, local chipping, or chip space reduction. When relying solely on image edges or surface brightness differences for judgment, only visible changes in shape are often observed, making it difficult to determine whether the surface microstructure has affected cutting performance. Furthermore, grayscale fluctuations in two-dimensional images are significantly affected by illumination angle, surface reflection state, and local texture distribution. Without coordinated constraints with three-dimensional coordinate information, local brightness changes can easily be mistaken for surface defects or roughness variations, thus affecting measurement consistency. Therefore, improvements are needed. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a laser scanning-based precision measurement system for diamond saw blade cutting edges.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a laser scanning-based diamond saw blade cutting edge precision measurement system includes: The speckle feature extraction module is used to control a single-wavelength laser beam to obliquely illuminate the cutting edge end face of the saw blade, call an industrial camera to collect the pixel matrix of corresponding bright and dark speckle images on the surface of the tire body, calculate and obtain the local speckle intensity comparison index, extract the preset surface scattering parameters and perform fitting calculation with the local speckle intensity comparison index to generate surface scattering mapping parameters. The roughness distribution construction module is used to calculate the root mean square roughness matrix of the end face based on the local speckle intensity comparison index and the surface scattering mapping parameter, input the mapping equation, construct the end face root mean square roughness matrix, control the laser scanner to perform line scanning operation along the cutting head area, capture the three-dimensional coordinate data points corresponding to the scanning beam, and perform coordinate registration and stitching of the three-dimensional coordinate data points with the end face root mean square roughness matrix to generate the cutting edge depth map matrix. The chip space calculation module is used to set the initial side length value to construct a spatial coverage cube unit, which covers the coordinate space corresponding to the cutting edge depth map matrix, and generates the spatial coverage cube unit distribution result. The spatial coverage statistics are repeatedly performed under different side length values ​​to construct a multi-scale spatial coverage dataset. The failure status assessment module is used to calculate and generate a morphological fractal dimension index based on the multi-scale spatial coverage dataset. If the morphological fractal dimension index is less than a preset decay period threshold, a wear failure command is issued to the terminal device.

[0006] Preferably, the step of obtaining the surface scattering mapping parameters is as follows: A single-wavelength laser beam is controlled to illuminate the cutting edge of the saw blade with a fixed oblique path. An industrial camera is used to continuously acquire images of corresponding bright and dark spots on the surface of the tire body. The pixel matrix of the bright and dark spot images is read point by point according to the pixel coordinates. The standard deviation and average intensity value of the intensity grayscale in the neighborhood of each pixel are statistically analyzed to form the pixel grayscale statistical results. Based on the pixel grayscale statistics, the intensity grayscale standard deviation and average intensity value corresponding to each pixel coordinate are extracted one by one. The intensity grayscale standard deviation value is divided by the average intensity value, and the corresponding ratios are written in the original pixel coordinate order. The ratio correspondence of all pixel coordinates is checked to obtain the local speckle intensity comparison index. The parameters corresponding to the oblique illumination state in the preset surface scattering parameters are retrieved, and the local speckle intensity comparison index is matched point by point to the preset surface scattering parameters according to the pixel coordinates. The correspondence is checked, the fitting results are screened, and the mapping values ​​are written to form the surface scattering mapping parameters.

[0007] Preferably, the step of obtaining the root mean square roughness matrix of the end face is as follows: Based on the local speckle intensity comparison index and the surface scattering mapping parameter, the corresponding values ​​of the local speckle intensity comparison index and the corresponding values ​​of the surface scattering mapping parameter are read point by point in the same spatial coordinate order. The corresponding values ​​of the local speckle intensity comparison index and the corresponding values ​​of the surface scattering mapping parameter are substituted into the corresponding input positions of the mapping equation, and the calculation is performed point by point. The calculation result corresponding to each spatial coordinate is recorded, and the calculation result corresponding to each spatial coordinate is defined as the root mean square roughness value of the surface region, forming a root mean square roughness value group of the surface region. Based on the root mean square roughness values ​​of the surface regions, extract the spatial coordinates bound to the root mean square roughness values ​​of each surface region, arrange them in order according to the row and column positions in the spatial coordinates, and write the root mean square roughness values ​​of each surface region into the corresponding matrix positions to form the end face root mean square roughness matrix.

[0008] Preferably, the steps for obtaining the cutting edge depth map matrix are as follows: The laser scanner is controlled to output line scanning trajectories along the cutting head area one by one. The three-dimensional coordinate data points corresponding to the scanning beam are continuously captured according to the line scanning trajectory. The spatial coordinates and depth positions of each three-dimensional coordinate data point are extracted. The spatial coordinates of the three-dimensional coordinate data points are mapped to the corresponding matrix positions of the root mean square roughness matrix of the end face to form a cutting edge depth map matrix.

[0009] Preferably, the steps for obtaining the spatial coverage cube unit distribution results are as follows: Read the spatial coordinates and depth positions corresponding to all matrix positions in the cutting edge depth map matrix, set the initial side length value, and divide the spatial covering cube units continuously along the three coordinate directions in the coordinate space corresponding to the cutting edge depth map matrix according to the initial side length value. Sequentially cover the corresponding coordinate area with each spatial covering cube unit, determine whether each three-dimensional data point falls into each spatial covering cube unit, record the position of the spatial covering cube unit that falls into the three-dimensional data point, count the number of spatial covering cube units with three-dimensional data points, and form the spatial covering cube unit distribution result.

[0010] Preferably, the steps for obtaining the multi-scale spatial coverage dataset are as follows: Extract the initial side length value and the number of spatial coverage cube units corresponding to the spatial coverage cube unit distribution result. Adjust the initial side length value according to the preset reduction step size. For each adjusted side length value, redivide the spatial coverage cube units. Re-execute the three-dimensional data point falling judgment and spatial coverage cube unit number statistics. Record the number of spatial coverage cube units corresponding to each side length value item by item. Write the correspondence between the side length value and the number of spatial coverage cube units in the same recording order to form the side length coverage pairing result. Extract all side length values ​​and all spatial coverage cube unit numbers from the side length coverage pairing results. Check each side length value for a unique corresponding spatial coverage cube unit number. Delete records with missing side length values ​​and records with missing spatial coverage cube unit numbers. Organize the remaining records continuously according to the order of the side length values. Write the organized side length values ​​and organized spatial coverage cube unit numbers into groups to form a multi-scale spatial coverage dataset.

[0011] Preferably, the step of obtaining the morphological fractal dimension index is as follows: Based on the multi-scale spatial cover dataset, each group of side length values ​​and the number of spatial cover cube units in each group are read item by item. Records with an empty number of spatial cover cube units are removed. The side length values ​​in the retained records are converted into logarithmic side length variables, and the number of spatial cover cube units in the retained records is converted into logarithmic unit number variables, forming a logarithmic variable pairing result. Based on the pairing results of the logarithmic variables, the fractal dimension index of the morphology is calculated.

[0012] Preferably, the step of obtaining the wear failure command is as follows: Extract the preset decay period threshold, compare the size relationship between the morphological fractal dimension index and the preset decay period threshold item by item. If the morphological fractal dimension index is less than the preset decay period threshold, write the wear state identifier, write the trigger time identifier and write the cutting edge position identifier, encapsulate the instruction content according to the preset instruction format and execute the sending, and issue the wear failure instruction to the terminal device.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, a pixel matrix of bright and dark speckle images is obtained by oblique illumination with a single-wavelength laser beam. A local speckle intensity comparison index is extracted from the standard deviation and average intensity values ​​of the pixel intensity grayscale. This local speckle intensity comparison index is then fitted with preset surface scattering parameters. This transforms the surface response, which was previously confined to the image grayscale level, into surface scattering mapping parameters that can participate in subsequent calculations, providing a quantitative basis for expressing the microscopic scattering differences on the cutting edge surface. Furthermore, the local speckle intensity comparison index and the surface scattering mapping parameters are input into a mapping equation to form a root mean square roughness matrix of the end face. Coordinate registration and stitching are then performed between the three-dimensional coordinate data points and the root mean square roughness matrix of the end face, simultaneously locking the surface roughness and spatial position relationship. This ensures that the cutting edge depth map matrix no longer only reflects the outer wheel. Instead of simply measuring the elevation, it simultaneously carries surface condition information and three-dimensional structural information, improving the accuracy of identifying local defects, fine wear, and edge contour restoration of the cutting edge. Further, it performs spatial coverage statistics under multi-sided conditions on the coordinate space of the cutting edge depth map matrix, forming a multi-scale spatial coverage dataset. Based on this, it calculates the fractal dimension index of the morphology and compares it with a preset decay period threshold. This allows the static morphology results to be advanced to the level of wear failure judgment, providing a unified measurement basis for changes in chip space, cutting edge complexity, and decay trend. This enhances the correlation between measurement results and actual usage conditions, as well as the direct decision-making value of the output results. It helps reduce biased judgments based solely on a single image observation and improves the comprehensive evaluation capability of the cutting edge's accuracy, wear degree, and failure timing. Attached Figure Description

[0014] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a three-dimensional distribution diagram of the root mean square roughness matrix of the end face; Figure 3 Paired fitting plots of logarithmic variables covering multi-scale spatial coverage; Figure 4 The image shows the wear and decay curve of the fractal dimension index of the morphology. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0016] Please see Figure 1-4 The present invention provides a technical solution: a laser scanning-based diamond saw blade cutting edge precision measurement system comprising: The speckle feature extraction module is used to control a single-wavelength laser beam to obliquely illuminate the cutting edge of the saw blade, call an industrial camera to collect the pixel matrix of corresponding bright and dark speckle images on the surface of the tire body, calculate and obtain the local speckle intensity comparison index, extract the preset surface scattering parameters and perform fitting calculation with the local speckle intensity comparison index to generate surface scattering mapping parameters. The roughness distribution construction module is used to construct the root mean square roughness matrix of the end face by inputting the mapping equation based on the local speckle intensity comparison index and surface scattering mapping parameters, controlling the laser scanner to perform line scanning operation along the cutting head area, capturing the three-dimensional coordinate data points corresponding to the scanning beam, and performing coordinate registration and stitching of the three-dimensional coordinate data points and the root mean square roughness matrix of the end face to generate the cutting edge depth map matrix. The chip space calculation module is used to set the initial side length value to construct a spatial coverage cube unit, which covers the coordinate space corresponding to the cutting edge depth map matrix, and generates the spatial coverage cube unit distribution result. The spatial coverage statistics are repeated under different side length values ​​to construct a multi-scale spatial coverage dataset. The failure status assessment module is used to calculate and generate a morphological fractal dimension index based on a multi-scale spatial coverage dataset. If the morphological fractal dimension index is less than a preset decay period threshold, a wear failure command is issued to the terminal device.

[0017] The steps for obtaining surface scattering mapping parameters are as follows: A single-wavelength laser beam is controlled to illuminate the cutting edge of the saw blade with a fixed oblique path. An industrial camera is used to continuously acquire images of corresponding bright and dark spots on the surface of the tire body. The pixel matrix of the bright and dark spot images is read point by point according to the pixel coordinates. The standard deviation and average intensity value of the intensity grayscale in the neighborhood of each pixel are statistically analyzed to form the pixel grayscale statistical results. Based on the pixel grayscale statistics, the intensity grayscale standard deviation and average intensity value corresponding to each pixel coordinate are extracted one by one. The intensity grayscale standard deviation value is divided by the average intensity value, and the corresponding ratios are written in the original pixel coordinate order. The ratio correspondence of all pixel coordinates is checked to obtain the local speckle intensity comparison index. The parameters corresponding to the oblique illumination state in the preset surface scattering parameters are retrieved. The local speckle intensity comparison index is matched to the preset surface scattering parameters point by point according to the pixel coordinates. The correspondence is checked, the fitting results are screened, and the mapping values ​​are written to form the surface scattering mapping parameters.

[0018] Specifically, a single-wavelength laser beam is controlled to illuminate the end face of the saw blade cutting edge with a fixed oblique path. An industrial camera continuously acquires images of corresponding bright and dark spots on the surface of the tire body. The wavelength of the single-wavelength laser beam is set to 650nm, the power is 5mW, and it is irradiated onto the end face of the tire body to be tested on the saw blade cutting edge at a fixed tilt angle of 45 degrees. The industrial camera uses a CMOS sensor with a resolution of 1920×1080 pixels and continuously captures speckle images formed by laser irradiation at a rate of 100 frames per second. Each frame of the acquired image is converted into an 8-bit grayscale image, resulting in a 2D pixel matrix of bright and dark spots. Then, each pixel in the pixel matrix is ​​traversed point by point in the pixel matrix in the order from left to right and from top to bottom. For any pixel with coordinates (i, j), its neighborhood is defined as a 7×7 pixel window centered on that point, that is, containing pixels from (i-3, j) to (i-1, j-1). For all pixels from (j-3) to (i+3,j+3), when processing pixels at the image edge, if the neighborhood window exceeds the image boundary, mirror filling is used to fill in the neighborhood data. Then, the gray values ​​of all 49 pixels in the 7×7 neighborhood are extracted, and the average value of these gray values ​​is calculated and recorded as the average intensity value of the pixel. At the same time, the standard deviation of these 49 gray values ​​is calculated and recorded as the intensity gray standard deviation value of the pixel. Each pixel coordinate (i, j) and its corresponding average intensity value and intensity gray standard deviation value are bound and stored. After traversing all pixels, a pixel gray-level statistical result containing all pixel coordinates, the neighborhood average intensity and neighborhood intensity standard deviation corresponding to each coordinate is formed.

[0019] Based on the pixel grayscale statistics, a 1920×1080 floating-point matrix with the same size as the original image is initialized to store the ratios for subsequent calculations. Then, the standard deviation and average intensity values ​​of each pixel are extracted. Before performing the division operation, an average intensity validity threshold is set. This threshold is calculated by adding three times the standard deviation to the average grayscale value of the dark-field image captured by the camera under no-light conditions. For example, if the average grayscale value in the dark field is 5 and the standard deviation is 1.5, then the threshold is set to 5 + 3 × 1.5 = 9.5. For each pixel, determine if its average intensity value is greater than the threshold 9.5. If it is less than or equal to the threshold, the pixel is considered to be an overly dark area or an invalid reflective point, and its speckle contrast information is unreliable. Write 0 directly to the ratio matrix position corresponding to the pixel coordinates. If the average intensity value is greater than the threshold, divide the intensity grayscale standard deviation of the pixel by its average intensity value to obtain a ratio, and store this ratio in the corresponding pixel coordinate position of the ratio matrix. After calculating for all pixels, check the integrity of the generated ratio matrix. This mainly confirms that each position in the matrix has been assigned a value (valid calculated value or default value 0), and there are no null values ​​or non-numeric (NaN) anomalies. This check process is automatically completed by a verification program. This program counts the number of valid values ​​in the matrix and compares them with the total number of pixels greater than the threshold to ensure consistency, thus guaranteeing a one-to-one correspondence of the data. The final filtered and verified ratio matrix is ​​the local speckle intensity contrast index.

[0020] The system retrieves parameters corresponding to the oblique illumination state from the preset surface scattering parameters. These parameters are not single values, but rather mapping models established through prior calibration experiments. The model is established as follows: A set of standard samples with nationally certified roughness values ​​is selected, for example, five samples with Ra values ​​of 0.2μm, 0.4μm, 0.8μm, 1.6μm, and 3.2μm. Each sample is measured using the system under the same 45-degree oblique illumination state. The average local speckle intensity comparison index of each sample surface is calculated, resulting in a set of data pairs (average local speckle intensity comparison index, standard roughness Ra value), such as (0.15, 0.2), (0.28, 0.4), (0.45, 0.8), (0.68, 1.6), and (0.85, 3.2). Polynomial fitting is then performed on these data pairs to obtain a second-order polynomial equation as the mapping model, for example, Ra = 1.2 × C. 2+ 2.5 × C + 0.05, where C is the local speckle intensity comparison index, Ra is the surface roughness, and the coefficients 1.2, 2.5, and 0.05 are the preset surface scattering parameters. Then, each value in the local speckle intensity comparison index matrix obtained in the previous step is substituted into this polynomial equation for calculation to generate an initial surface roughness distribution matrix. Next, the fitting results are screened. First, a 3×3 median filter is applied to traverse the entire initial surface roughness distribution matrix, and the original value is replaced with the roughness median in the neighborhood of each point to remove sudden noise points. Then, the average roughness and standard deviation of the entire median-filtered matrix are calculated, and a screening interval is set as the average value plus or minus three times the standard deviation. The matrix is ​​traversed again, and the values ​​outside this interval are replaced with the average of the effective values ​​in their 3×3 neighborhood (excluding themselves). After verification and screening, the finally obtained stable and reliable roughness data matrix is ​​written and defined as the surface scattering mapping parameter.

[0021] The steps for obtaining the root mean square roughness matrix of the end face are as follows: Based on the local speckle intensity comparison index and the surface scattering mapping parameter, the corresponding values ​​of the local speckle intensity comparison index and the corresponding values ​​of the surface scattering mapping parameter are read point by point in the same spatial coordinate order. The corresponding values ​​of the local speckle intensity comparison index and the corresponding values ​​of the surface scattering mapping parameter are substituted into the corresponding input positions of the mapping equation, and the calculation is performed point by point. The calculation result corresponding to each spatial coordinate is recorded, and the calculation result corresponding to each spatial coordinate is defined as the root mean square roughness value of the surface region, forming a root mean square roughness value group of the surface region. Based on the root mean square roughness values ​​of the surface regions, the spatial coordinates bound to the root mean square roughness values ​​of each surface region are extracted. The coordinates are then arranged in order according to the row and column positions in the spatial coordinates. The root mean square roughness values ​​of each surface region are written into the corresponding matrix positions to form the end face root mean square roughness matrix.

[0022] Specifically, based on the local speckle intensity contrast index and the surface scattering mapping parameter, a point-by-point calculation process is initiated. This process first reads a value from the local speckle intensity contrast index matrix and the surface scattering mapping parameter from memory in the same spatial coordinate order. The mapping equation is a second-order polynomial model determined during the previous calibration process, specifically in the form of... ,in, In coordinates The calculated root mean square roughness value for the surface region is given. From the local speckle intensity comparison index matrix, at coordinates The value read from the location, coefficient This refers to the surface scattering mapping parameter, whose value is obtained by fitting the measurement results of different standard roughness samples, for example... , , This operation is performed within a loop, where the loop variable iterates through the spatial coordinates of all pixels, starting from... Start to End, in each loop, change the current coordinates Corresponding local speckle intensity comparison index value Substitute into the mapping equation above to calculate the point. Value, for example, if a certain point of If the value is 0.5, then its root mean square roughness value is... And compare this calculation result 1.6 with its spatial coordinates. The data is bound together and stored as a data pair in a temporary list. After traversing all pixels and completing the calculation, the temporary list contains all spatial coordinates and their corresponding root mean square roughness values, forming a group of root mean square roughness values ​​for the surface region.

[0023] Based on the root mean square roughness values ​​of the surface region, the dimensions of the target matrix are first determined. This is done by reading the maximum values ​​of the row and column positions of all spatial coordinates in the root mean square roughness values ​​(e.g., 1920 and 1080), creating a 1920×1080 two-dimensional array stored as floating-point numbers. This array is then initialized with an invalid identifier (e.g., -1.0). Next, each data pair in the root mean square roughness values ​​is iterated through. For each data pair, its associated spatial coordinates, such as the row position, are extracted. and column position And the corresponding root mean square roughness value of the surface area. Then the value Write to the newly created two-dimensional array The initial invalid identifier -1.0 is replaced at the location. This process essentially rearranges the data in the unordered list into a structured two-dimensional matrix according to the natural order of spatial coordinates. After all data pairs in the root mean square roughness value group of the surface area have been processed and written, an integrity check is performed. This check traverses the entire newly generated two-dimensional array and counts the number of elements with a value of -1.0. If there is an element with a value of -1.0, it means that the data at that location is missing in the previous calculation. At this time, a filling and repair procedure is started. For each missing point, the average of the effective roughness values ​​(i.e., values ​​that are not -1.0) of its 8 neighboring points in a 3×3 neighborhood is used for filling. If all 8 neighboring points are -1.0, the search range is expanded to a 5×5 neighborhood until an effective neighboring point is found or the preset maximum search radius of 10 pixels is reached. After filling all missing points, the final complete and continuous two-dimensional array is the root mean square roughness matrix of the end face.

[0024] The steps to obtain the cutting edge depth map matrix are as follows: The laser scanner is controlled to output line scanning trajectories along the cutting head area one by one. The three-dimensional coordinate data points corresponding to the scanning beam are continuously captured according to the line scanning trajectory. The spatial coordinates and depth positions of each three-dimensional coordinate data point are extracted. The spatial coordinates of the three-dimensional coordinate data points are mapped to the corresponding matrix positions of the root mean square roughness matrix of the end face to form the cutting edge depth map matrix.

[0025] Specifically, the laser scanner is controlled to output line scanning trajectories along the blade head area one by one. The scanning speed of the laser scanner is set to 50 millimeters per second, the line scanning frequency is 2 kHz, and the spacing between adjacent scanning lines is controlled at 0.01 millimeters to ensure high-density data acquisition of the blade head area. During the scanning process, three-dimensional coordinate data points formed by the reflection of the scanning beam on the tire surface are continuously captured. Each data point contains its coordinates in the scanner coordinate system. coordinates, where Representing depth position, after data acquisition, the scanner's physical coordinate system needs to be registered with the pixel coordinate system of the industrial camera image. This registration is achieved through a pre-calculated affine transformation matrix. This matrix is ​​achieved during the equipment calibration phase by photographing and scanning a calibration board with at least four non-collinear marker points. The registration process is as follows: traversing each acquired 3D coordinate data point... Applying the inverse of the affine transformation matrix Its Coordinates to pixel coordinates ,Right now Due to the calculated It's a floating-point number; round it to the nearest integer pixel coordinate. Then, create a depth matrix with the same size as the root mean square roughness matrix of the end face (e.g., 1920×1080) and initialize all its elements to zero, thus setting the depth position of the 3D data points. Assigned to the depth matrix At the corresponding location, because the scanning resolution may be higher than the camera resolution, multiple 3D points may be mapped to the same pixel coordinate. In this case, all points are mapped to the same pixel. depth value The depth values ​​are accumulated and the number of points mapped to each pixel is recorded. After all 3D points have been processed, the accumulated depth value at each position in the depth matrix is ​​divided by the corresponding number of points to obtain the average depth value. The final matrix that stores the average depth of each pixel is the cutting edge depth map matrix.

[0026] The steps for obtaining the spatial coverage cube cell distribution results are as follows: Read the spatial coordinates and depth positions corresponding to all matrix positions in the cutting edge depth map matrix, set the initial side length value, and continuously divide the spatial covering cube units along three coordinate directions in the coordinate space corresponding to the cutting edge depth map matrix according to the initial side length value. Sequentially cover the corresponding coordinate area with each spatial covering cube unit, determine whether each 3D data point falls into each spatial covering cube unit, record the position of the spatial covering cube unit that falls into the 3D data point, count the number of spatial covering cube units with 3D data points, and form the spatial covering cube unit distribution result.

[0027] Specifically, the spatial coordinates and depth positions corresponding to all matrix positions in the cutting edge depth map matrix are read. First, the two-dimensional pixel coordinates in the cutting edge depth map matrix are... and depth value Converted into three-dimensional physical space coordinates through a preset affine transformation matrix. A 3D point cloud is formed. The maximum and minimum values ​​of this point cloud along the three coordinate axes are then calculated. A minimum orthogonal rectangular bounding box that completely encloses all data points is determined, and an initial side length is set. This initial side length is based on one-quarter of the maximum span of the bounding box in the three dimensions. For example, if the spans of the point cloud in the x, y, and z directions are 20mm, 30mm, and 5mm respectively, then the maximum span is 30mm, and the initial side length is set to 7.5mm. Based on this initial side length of 7.5mm, starting from the minimum coordinate point of the bounding box within the coordinate space of the point cloud, along the x, y, z axes... A grid with a side length of 7.5mm is continuously divided in three directions (z, z, z) to generate a series of non-overlapping spatial covering cube cells. Next, a Boolean array with the same dimensions as the 3D grid is created, and all elements are initialized to false. Then, each data point in the 3D data point cloud is traversed, and the index of the spatial covering cube cell to which the point belongs is calculated by dividing the coordinate value of the point by the side length and rounding down. The element value at the corresponding index position in the Boolean array is set to true, indicating that at least one 3D data point falls within the cube cell. After traversing all 3D data points, the total number of elements with a true value in the Boolean array is counted. This total number is the number of spatial covering cube cells containing 3D data points, thus forming the spatial covering cube cell distribution result containing the initial side length value and the corresponding number of cells.

[0028] The steps for obtaining a multi-scale spatial coverage dataset are as follows: Extract the initial side length value and the number of spatial covering cube units corresponding to the spatial covering cube unit distribution results. Adjust the initial side length value according to the preset reduction step size. For each adjusted side length value, redivide the spatial covering cube units. Re-execute the 3D data point falling judgment and spatial covering cube unit number statistics. Record the number of spatial covering cube units corresponding to each side length value item by item. Write the correspondence between the side length value and the number of spatial covering cube units in the same recording order to form the side length covering pairing result. Extract all side length values ​​and all spatial coverage cube cell counts from the side length coverage pairing results. Check each side length value for a unique corresponding spatial coverage cube cell count. Delete records with missing side length values ​​and records with missing spatial coverage cube cell counts. Organize the remaining records continuously according to the order of the side length values. Write the organized side length values ​​and organized spatial coverage cube cell counts into groups to form a multi-scale spatial coverage dataset.

[0029] Specifically, the initial side length value of 7.5mm corresponding to the spatial covering cube cell distribution result and the calculated number of spatial covering cube cells are extracted and recorded as the first set of data. Then, the initial side length value is reduced by a preset geometric reduction step size, which is set as a multiplication factor less than 1, such as 0.9. This factor is set empirically to strike a balance between ensuring a sufficient number of data points and computational efficiency. The current side length value is multiplied by 0.9 to obtain the new side length value; for example, after the first reduction, the side length is 7.5mm * 0.9 = The new side length is set to 6.75mm. Using this new side length, the bounding box of the entire 3D point cloud is re-meshed using 6.75mm as the side length. A new Boolean array with a larger dimension is created, and all 3D data points are iterated through to determine the new cube cell position they fall into. Finally, the number of spatial coverage cube cells occupied by the data points at the 6.75mm scale is counted. This new side length of 6.75mm and the newly counted number of cells are recorded as the second set of data. This process is repeated in a loop, with the previous side length value multiplied by 0.9 each time, and a complete spatial coverage count is performed. The loop terminates when the reduced side length is less than the measurement resolution of the laser scanner, for example, 0.01mm, because subdivisions smaller than this scale have no physical meaning. Throughout the loop, the side length value obtained in each iteration and the corresponding number of spatial coverage cube cells are written into a list according to the order of iteration, forming the side length coverage pairing results.

[0030] Extract all side length values ​​and the number of spatial covering cube units from the side length coverage pairing results. Initiate a data cleaning and processing process. This process iterates through each data pair in the side length coverage pairing results list, performing a validity check on each pair. The check conditions are that the side length value must be a positive number and the number of spatial covering cube units must also be a positive integer. Any record with a side length value of zero or negative, or a unit count of zero, is considered an invalid calculation result and is deleted from the list. For example, if a side length value becomes non-positive due to floating-point precision issues, or if the scale is too small and no cube is occupied, the record will be deleted. After deleting all invalid records, all valid data in the list are processed. According to the data, the data is sorted in descending order of side length values. This sorting step ensures data consistency and the monotonicity of subsequent logarithmic transformations. After sorting, the total number of records to be retained is checked again, and a minimum record number threshold is set. This threshold is set according to the minimum requirements for subsequent linear fitting, and is usually empirically set to 5, meaning that at least 5 data points at different scales are needed for effective fitting analysis. If the number of retained records is less than 5, the measurement data is considered insufficient and marked as an error. If the number of records is greater than or equal to 5, the cleaned and sorted side length values ​​and the corresponding number of spatial cover cube units are written as a two-dimensional array or structured list for final grouping, forming a multi-scale spatial cover dataset.

[0031] The steps for obtaining the fractal dimension index of morphology are as follows: Based on the multi-scale spatial cover dataset, the side length values ​​and the number of spatial cover cube units in each group are read item by item. Records with an empty number of spatial cover cube units are removed. The side length values ​​in the retained records are converted into logarithmic side length variables, and the number of spatial cover cube units in the retained records is converted into logarithmic unit number variables, forming a logarithmic variable pairing result. Based on the pairing results of logarithmic variables, the fractal dimension index of the morphology is calculated using the following formula: ; ; in, The variance of all logarithmic side length variables. This represents the total number of record groups in the logarithmic variable pairing results. The first in the pairing results of logarithmic variables Group record number, For the first The logarithmic side length variable corresponding to the group record The average of all logarithmic side length variables. , For the first The logarithmic number of units corresponding to the group record, The average of all logarithmic unit number variables. , It is the fractal dimension index of the morphology.

[0032] Specifically, based on the multi-scale spatial cover dataset, a data transformation process is initiated. This process reads each set of side length values ​​and the corresponding number of spatial cover cube cells in the dataset item by item. During the reading process, a data validity check is performed, discarding all records with a spatial cover cube cell count equal to or less than zero. For example, if a record is (6.75mm, 0), that record will be discarded. For all retained records that pass the validity check, the side length values ​​and spatial cover cube cell counts are batch-transformed logarithmically using the natural constant e as the basis. The logarithmic function at the bottom is used to calculate the logarithmic side length variable (ln(7.5)) approximately equal to 2.015 and the logarithmic number of units (ln(120)) approximately equal to 4.787 for each retained record, such as one containing a side length of 7.5 mm and 120 spatial cover cube units. This new logarithmic value (2.015, 4.787) is stored as a data point. This transformation process is applied to all valid data pairs in the multi-scale spatial cover dataset. For example, a multi-scale spatial cover dataset containing 5 sets of valid data has the following content: ((7.5, The values ​​(120), (6.75, 150), (6.075, 185), (5.4675, 230), (4.9208, 290)) are transformed one by one, resulting in a new list containing 5 pairs of logarithmic variables. The contents are approximately ((2.015, 4.787), (1.909, 5.011), (1.804, 5.220), (1.700, 5.438), (1.593, 5.670)). Each pair of values ​​in this list is a logarithmic variable pair. All these pairing results are then set in their original order to form the logarithmic variable pairing results.

[0033] In the formula for calculating the fractal dimension index of morphology, the variance of the logarithmic side length variable is introduced into the denominator. Each data point is weighted, taking into account the dispersion of the independent variable itself, giving greater influence to those off-center, more informative points, while correcting for overall fluctuations through the variance term. The calculated slope has better robustness to measurement noise and abnormal fluctuations of data points. This represents the total number of records in the logarithmic variable pairing results. For example, if the logarithmic variable pairing results contain the following 5 data sets: ((2.015, 4.787), (1.909, 5.011), (1.804, 5.220), (1.700, 5.438), (1.593, 5.670)), then... The process of obtaining the list involves counting the number of elements in the list. .

[0034] For the first The logarithmic side length variable corresponding to the group record is obtained directly from the pairing results of the logarithmic variable by ordinal number. Extraction represents the first The logarithm value of the i-th scale is obtained by reading the i-th value from the list of logarithmic variable pairing results. The first element of a tuple, for example, when At that time, it was extracted from the paired results of the logarithmic variable ((2.015, 4.787), ...). .

[0035] For the first The logarithmic unit number variable corresponding to the group record is also derived from the logarithmic variable pairing results by sequence number. Extraction represents the extraction in the first... The logarithm of the number of cubic units required to cover the surface at each scale is obtained by reading the logarithmic variable and matching it with the first result in the result list. The second element of a tuple, for example, when At that time, it was extracted from the paired results of the logarithmic variable ((2.015, 4.787), ...). .

[0036] The parameter is the average of all logarithmic side length variables, calculated by applying the average of all logarithmic side length variables. indivual The average of the values ​​is obtained, which reflects the center position of the selected measurement scale. The calculation process involves averaging all the values. Add the values ​​and then divide by For example, using the first 5 sets of data, .

[0037] The average value of the total number of logarithmic units is calculated by applying the average value of all logarithmic units. indivual The value is obtained by averaging, which reflects the logarithmic number of units required to cover the surface at an average scale. The calculation process involves averaging all the units... Add the values ​​and then divide by For example, using the first 5 sets of data, .

[0038] Calculations based on parameters: Based on the paired logarithmic variables ((2.015, 4.787), (1.909, 5.011), (1.804, 5.220), (1.700, 5.438), (1.593, 5.670)), we obtain... , , .

[0039] First calculate : ; Next, calculate The summation term of the numerator and denominator, with For example: Numerator ; denominator ; For all 5 points ( Calculate and sum the results in step 5, and finally calculate... : .

[0040] The results indicate that the measured fractal dimension of the saw blade cutting edge is 2.7969. A higher fractal dimension signifies a rich micro-geometry on the surface, corresponding to a large number of sharp cutting edges, a characteristic of a saw blade in good working condition. Conversely, if the calculated... A lower value, such as close to 2.0, indicates that the surface tends to be smooth and has lost its complex cutting structure, and may have already experienced severe wear. Therefore, the result of 2.7969 indicates that the current saw blade is in good condition and has high surface complexity.

[0041] The steps for obtaining the wear failure command are as follows: Extract the preset decay period threshold, compare the size relationship between the morphological fractal dimension index and the preset decay period threshold item by item. If the morphological fractal dimension index is less than the preset decay period threshold, write the wear status identifier, write the trigger time identifier and write the cutting edge position identifier, encapsulate the instruction content according to the preset instruction format and execute the sending, and issue the wear failure instruction to the terminal device.

[0042] Specifically, a preset decay period threshold is extracted. This threshold is not a fixed empirical value, but is calibrated through a full life-cycle wear test on new saw blades of the same model. The calibration process is as follows: Take at least 5 brand-new diamond saw blades. First, measure their initial fractal dimension index to obtain a set of initial values, such as [2.81, 2.79, 2.85, 2.82, 2.83]. Then, put these saw blades into a standard cutting test, cutting a fixed material such as granite. After cutting 50 meters, take out the saw blade and measure the fractal dimension index again, while simultaneously recording the cutting efficiency and workpiece surface quality. Continue this process until the saw blade's cutting efficiency drops below 60% of the initial efficiency, or the workpiece surface shows severe chipping or burning. This is defined as the saw blade entering the decay period. Record the last measurement of the fractal dimension index before entering the decay period. Through the experiment of 5 saw blades, a set of critical decay period index values ​​is obtained, such as [2.25, 2.28, 2.23, ...]. [2.26, 2.24], the average of this set of critical values ​​is calculated to be 2.252. To ensure the reliability of the early warning, the preset decay period threshold is set to this average value plus one standard deviation, i.e. Here, the average value is used as the threshold of 2.25. Then, the currently calculated fractal dimension index of 2.7969 is compared with the preset decay period threshold of 2.25. Since 2.7969 is greater than 2.25, the saw blade is judged to be in normal condition, and no operation is performed. If the fractal dimension index of 2.21 is obtained in a certain measurement, then 2.21 is less than 2.25, and the wear failure response is triggered. The wear status identifier is immediately set to 1 (representing failure), the system clock is called to obtain the current precise timestamp, such as "2025-10-26 15:08:33.128", and the saw blade number and cutter head area measured at the moment are recorded, such as "Blade03-Sector08". Subsequently, these three pieces of information (status identifier 1, timestamp, and location identifier) ​​are processed in a predefined JSON format ({"type":"alert", "code":"wear_fail", "value":1, "time":"2025-10-26"). The data packet (15:08:33.128", "source":"Blade03-Sector08"}) is encapsulated and sent to the terminal device in the central control room (IP address 192.168.1.100, port 9001) via TCP / IP protocol, issuing a wear and failure command.

Claims

1. A laser scanning-based precision measurement system for diamond saw blade cutting edges, characterized in that, The system includes: The speckle feature extraction module is used to control a single-wavelength laser beam to obliquely illuminate the cutting edge end face of the saw blade, call an industrial camera to collect the pixel matrix of corresponding bright and dark speckle images on the surface of the tire body, calculate and obtain the local speckle intensity comparison index, extract the preset surface scattering parameters and perform fitting calculation with the local speckle intensity comparison index to generate surface scattering mapping parameters. The roughness distribution construction module is used to calculate the root mean square roughness matrix of the end face based on the local speckle intensity comparison index and the surface scattering mapping parameter, input the mapping equation, construct the end face root mean square roughness matrix, control the laser scanner to perform line scanning operation along the cutting head area, capture the three-dimensional coordinate data points corresponding to the scanning beam, and perform coordinate registration and stitching of the three-dimensional coordinate data points with the end face root mean square roughness matrix to generate the cutting edge depth map matrix. The chip space calculation module is used to set the initial side length value to construct a spatial coverage cube unit, which covers the coordinate space corresponding to the cutting edge depth map matrix, and generates the spatial coverage cube unit distribution result. The spatial coverage statistics are repeatedly performed under different side length values ​​to construct a multi-scale spatial coverage dataset. The failure status assessment module is used to calculate and generate a morphological fractal dimension index based on the multi-scale spatial coverage dataset. If the morphological fractal dimension index is less than a preset decay period threshold, a wear failure command is issued to the terminal device.

2. The laser scanning-based diamond saw blade cutting edge accuracy measurement system according to claim 1, characterized in that, The steps for obtaining the surface scattering mapping parameters are as follows: A single-wavelength laser beam is controlled to illuminate the cutting edge of the saw blade with a fixed oblique path. An industrial camera is used to continuously acquire images of corresponding bright and dark spots on the surface of the tire body. The pixel matrix of the bright and dark spot images is read point by point according to the pixel coordinates. The standard deviation and average intensity value of the intensity grayscale in the neighborhood of each pixel are statistically analyzed to form the pixel grayscale statistical results. Based on the pixel grayscale statistics, the intensity grayscale standard deviation and average intensity value corresponding to each pixel coordinate are extracted one by one. The intensity grayscale standard deviation value is divided by the average intensity value, and the corresponding ratios are written in the original pixel coordinate order. The ratio correspondence of all pixel coordinates is checked to obtain the local speckle intensity comparison index. The parameters corresponding to the oblique illumination state in the preset surface scattering parameters are retrieved, and the local speckle intensity comparison index is matched point by point to the preset surface scattering parameters according to the pixel coordinates. The correspondence is checked, the fitting results are screened, and the mapping values ​​are written to form the surface scattering mapping parameters.

3. The laser scanning-based diamond saw blade cutting edge accuracy measurement system according to claim 1, characterized in that, The steps for obtaining the root mean square roughness matrix of the end face are as follows: Based on the local speckle intensity comparison index and the surface scattering mapping parameter, the corresponding values ​​of the local speckle intensity comparison index and the corresponding values ​​of the surface scattering mapping parameter are read point by point in the same spatial coordinate order. The corresponding values ​​of the local speckle intensity comparison index and the corresponding values ​​of the surface scattering mapping parameter are substituted into the corresponding input positions of the mapping equation, and the calculation is performed point by point. The calculation result corresponding to each spatial coordinate is recorded, and the calculation result corresponding to each spatial coordinate is defined as the root mean square roughness value of the surface region, forming a root mean square roughness value group of the surface region. Based on the root mean square roughness values ​​of the surface regions, extract the spatial coordinates bound to the root mean square roughness values ​​of each surface region, arrange them in order according to the row and column positions in the spatial coordinates, and write the root mean square roughness values ​​of each surface region into the corresponding matrix positions to form the end face root mean square roughness matrix.

4. The laser scanning-based diamond saw blade cutting edge accuracy measurement system according to claim 1, characterized in that, The steps for obtaining the cutting edge depth map matrix are as follows: The laser scanner is controlled to output line scanning trajectories along the cutting head area one by one. The three-dimensional coordinate data points corresponding to the scanning beam are continuously captured according to the line scanning trajectory. The spatial coordinates and depth positions of each three-dimensional coordinate data point are extracted. The spatial coordinates of the three-dimensional coordinate data points are mapped to the corresponding matrix positions of the root mean square roughness matrix of the end face to form a cutting edge depth map matrix.

5. The laser scanning-based diamond saw blade cutting edge accuracy measurement system according to claim 1, characterized in that, The steps for obtaining the spatial coverage cube unit distribution results are as follows: Read the spatial coordinates and depth positions corresponding to all matrix positions in the cutting edge depth map matrix, set the initial side length value, and divide the spatial covering cube units continuously along the three coordinate directions in the coordinate space corresponding to the cutting edge depth map matrix according to the initial side length value. Sequentially cover the corresponding coordinate area with each spatial covering cube unit, determine whether each three-dimensional data point falls into each spatial covering cube unit, record the position of the spatial covering cube unit that falls into the three-dimensional data point, count the number of spatial covering cube units with three-dimensional data points, and form the spatial covering cube unit distribution result.

6. The laser scanning-based diamond saw blade cutting edge accuracy measurement system according to claim 1, characterized in that, The steps for obtaining the multi-scale spatial coverage dataset are as follows: Extract the initial side length value and the number of spatial coverage cube units corresponding to the spatial coverage cube unit distribution result. Adjust the initial side length value according to the preset reduction step size. For each adjusted side length value, redivide the spatial coverage cube units. Re-execute the three-dimensional data point falling judgment and spatial coverage cube unit number statistics. Record the number of spatial coverage cube units corresponding to each side length value item by item. Write the correspondence between the side length value and the number of spatial coverage cube units in the same recording order to form the side length coverage pairing result. Extract all side length values ​​and all spatial coverage cube unit numbers from the side length coverage pairing results. Check each side length value for a unique corresponding spatial coverage cube unit number. Delete records with missing side length values ​​and records with missing spatial coverage cube unit numbers. Organize the remaining records continuously according to the order of the side length values. Write the organized side length values ​​and organized spatial coverage cube unit numbers into groups to form a multi-scale spatial coverage dataset.

7. The laser scanning-based diamond saw blade cutting edge accuracy measurement system according to claim 1, characterized in that, The steps for obtaining the fractal dimension index of the morphology are as follows: Based on the multi-scale spatial cover dataset, each group of side length values ​​and the number of spatial cover cube units in each group are read item by item. Records with an empty number of spatial cover cube units are removed. The side length values ​​in the retained records are converted into logarithmic side length variables, and the number of spatial cover cube units in the retained records is converted into logarithmic unit number variables, forming a logarithmic variable pairing result. Based on the pairing results of the logarithmic variables, the fractal dimension index of the morphology is calculated.

8. The laser scanning-based diamond saw blade cutting edge accuracy measurement system according to claim 1, characterized in that, The steps for obtaining the wear failure command are as follows: Extract the preset decay period threshold, compare the size relationship between the morphological fractal dimension index and the preset decay period threshold item by item. If the morphological fractal dimension index is less than the preset decay period threshold, write the wear state identifier, write the trigger time identifier and write the cutting edge position identifier, encapsulate the instruction content according to the preset instruction format and execute the sending, and issue the wear failure instruction to the terminal device.

Images