Powder grinding quality detection method and system

Abstract

The invention relates to the technical field of image processing, in particular to a powder grinding quality detection method and system. The method comprises the following steps: acquiring a powder image, and segmenting the powder image to form a plurality of image blocks; constructing a gray feature vector of each image block, and performing clustering processing on each image block to obtain a plurality of classification clusters; calculating the upper-layer probability that each cluster is an upper-layer ore, and determining an upper adjustment coefficient and a lower adjustment coefficient of each cluster; and determining an optimal upper threshold value and an optimal lower threshold value of a canny operator based on the upper adjustment coefficient and the lower adjustment coefficient, carrying out edge detection on the image blocks in each cluster by using the canny operator, and judging whether the powder is qualified or not. The ore particle detection method has the effect of improving the ore particle detection accuracy.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a powder grinding quality detection method and system. Background Art

[0002] In the ore processing industry, in order to extract the valuable components in the ore, the ore needs to be crushed and ground. Only when the ore particle size reaches a certain level can the effective components in the ore be fully extracted. Therefore, the particle size of the ground ore powder is an important indicator in the grinding process. In order to ensure the quality of the ore powder grinding, the ore powder needs to be tested during the grinding process. The traditional detection method is the laboratory sampling detection method, which requires multiple manual sampling and testing of the samples in the laboratory. On the one hand, this method consumes a lot of human resources, and on the other hand, the detection efficiency is also low. Therefore, image processing detection technology is gradually applied to the process of ore powder grinding. For example, a method for evaluating the ore crushing effect based on image processing is disclosed in the patent application document with publication number CN116823827A; it mainly includes the following steps: obtaining a grayscale image of the ore, generating an ore shape image through a particle detection algorithm, segmenting the ore shape image, generating an ore particle image, and then evaluating the ore crushing effect.

[0003] The common Canny edge detection algorithm can be applied to the extraction step of ore shape images. The Canny algorithm mainly includes the following steps: calculating pixel gradients, presetting high thresholds and low thresholds; and screening pixels according to pixel gradients and thresholds. In the process of ore detection, ore materials are usually stacked. According to the position of ore particles, ore particles can be divided into upper ore on the surface of the material pile and lower ore particles in the lower layer. Ore particles in the lower layer are usually affected by upper particles. For example, the light is blocked by the upper ore, resulting in a small change in the grayscale of the edge of the lower ore particles, which in turn leads to the loss of the edge of the lower ore particles, and ultimately leads to low accuracy in ore particle detection. Summary of the invention

[0004] In order to solve the problem of low accuracy in ore particle detection, the present application provides a powder grinding quality detection method and system.

[0005] In the first aspect, the present application provides a method for detecting the grinding quality of powder material, which adopts the following technical solution: A powder grinding quality detection method comprises the following steps: collecting a powder image, segmenting the powder image to form a plurality of image blocks; constructing a grayscale feature vector of each image block, performing clustering processing on each image block to obtain a plurality of classification clusters; calculating the upper layer probability of each cluster being an upper layer ore, and determining an upper adjustment coefficient and a lower adjustment coefficient of each cluster; determining an optimal upper threshold and an optimal lower threshold of a Canny operator based on the upper adjustment coefficient and the lower adjustment coefficient, and performing edge detection on the image blocks in each cluster using the Canny operator to judge whether the powder is qualified; The calculation formula for the upper probability of clustering is: ; In the formula, Indicates The upper probability of clusters; Indicates The mean of the overall grayscale of all blocks in a cluster; Indicates The mean of the global deviations of all tiles in the cluster; Indicates The mean internal contrast of all patches in a cluster; Expressed as a natural constant is the logarithmic function of the base; is a linear normalization function.

[0006] First, the powder image is divided into multiple blocks, and then the blocks are clustered to form clusters with similar features. The probability that each cluster belongs to the upper layer of ore is calculated, and the upper threshold of the canny operator is adjusted by the probability that each cluster belongs to the upper layer of ore to obtain the optimal upper threshold, and the lower threshold is adjusted to obtain the optimal lower threshold. The upper layer of ore usually has the characteristic of large edge gradient amplitude, so the upper threshold of the canny operator can be increased when using canny for edge extraction for the upper layer of ore. The amplitude of the edge of the lower layer of ore is reduced relative to the amplitude of the edge of the upper layer of ore due to the influence of the upper layer of ore, so the conditions for edge pixel screening can be relaxed. In this application, the upper and lower thresholds in the process of extracting the edge of the canny operator are dynamically adjusted according to the upper layer probability corresponding to each cluster, which can reduce the situation where the high saturation area in the powder image is misjudged as an edge pixel on the one hand, and the edge of the corresponding lower layer of ore can automatically relax the edge pixel screening conditions, thereby reducing the loss of the edge of the lower layer of ore and improving the accuracy of ore particle detection.

[0007] Optionally, the step of calculating the overall grayscale of the image block includes obtaining the grayscale value of each pixel in the image block; and taking the average of the grayscale values ​​of all the pixels in the image block as the overall grayscale of the image block.

[0008] The average grayscale value of all pixels in the block is calculated to obtain the overall grayscale value that can reflect the grayscale of the entire block.

[0009] Optionally, the internal contrast of the tile is calculated as: ; Indicates Internal contrast in each tile; Indicates The total number of pixels in a tile; Indicates the pixel ordinal number; Indicates In the block The gray value of each pixel; Indicates The mean gray value of all pixels in a block; Represents the standard normalization function.

[0010] It indicates the difference between any pixel in the block and the overall grayscale of the block. The larger the difference, the more discrete the grayscale data of each pixel in the block. The upper ore has clear edges and drastic changes in light and dark. If the block is the upper ore, the grayscale value of each pixel in the block is relatively discrete. Therefore, the internal contrast of the block calculated in this formula helps to distinguish whether the block is in the upper ore.

[0011] Optionally, the internal contrast of the tile is calculated as: ; Indicates Internal contrast in each tile; Indicates The total number of pixels in a tile; Indicates the pixel ordinal number; Indicates In the block The gray value of each pixel; Indicates The mean gray value of all pixels in a block; Represents the standard normalization function.

[0012] This formula introduces absolute value calculation, which is relatively simple and efficient, and can be applied to environments with high requirements for calculation efficiency.

[0013] Optionally, the step of calculating the global deviation of the image block includes: obtaining the grayscale values ​​of all pixels of the powder image and calculating the average value to obtain the global grayscale; and taking the difference between the overall grayscale of the image block and the global grayscale as the global deviation.

[0014] Optionally, the feature vector includes: the overall grayscale of the tile, the global grayscale of the tile, and the internal contrast of the tile.

[0015] The eigenvalues ​​of multiple dimensions together constitute the eigenvector, which is used as a metric in the subsequent clustering process to comprehensively consider the characteristics of multiple faces of the block, thereby improving the accuracy of the clustering results and providing effective data support for subsequent calculations, which in turn helps to improve the accuracy of edge detection results.

[0016] Optionally, an iterative self-organizing clustering algorithm is used to cluster the tiles using the feature vectors as a metric.

[0017] Optionally, the powder image is segmented using a superpixel segmentation method.

[0018] Optionally, in the step of calculating the upper adjustment coefficient and the lower adjustment coefficient of each cluster according to the upper probability of each cluster, the calculation formula of the upper adjustment coefficient is: , where Indicates The up-regulation coefficient of each cluster; Indicates The upper probability of clusters; The calculation formula of the lower adjustment coefficient is: , where Indicates The down-regulation coefficient of the clusters; Indicates The upper probability of the clusters.

[0019] The upper adjustment coefficient and the lower adjustment coefficient are determined according to the upper layer probability. When the upper layer probability changes, the upper adjustment coefficient and the lower adjustment coefficient calculated in the formula change accordingly, so that the upper threshold and the lower threshold of the Canny algorithm can be dynamically adjusted.

[0020] In the second aspect, the present application provides a powder grinding quality detection system, which adopts the following technical solution: A powder grinding quality detection system comprises a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a powder grinding quality detection method according to the above is implemented.

[0021] The above-mentioned powder grinding quality detection method is generated into a computer program and stored in a memory so as to be loaded and executed by a processor. Thus, a system is made based on the memory and the processor for easy use.

[0022] The present application has the following technical effects: the powder image is divided into multiple blocks, and the blocks are clustered. By calculating the upper layer probabilities of different clustering clusters, the upper adjustment coefficient and the lower adjustment coefficient suitable for each clustering cluster can be calculated. When the Canny algorithm extracts the edges of the blocks in each clustering cluster, the upper threshold of the Canny algorithm is adjusted in a targeted manner through the upper adjustment coefficient of each clustering cluster, and the lower threshold of the Canny algorithm is adjusted in a targeted manner according to the lower adjustment coefficient of each clustering cluster, so that a threshold that is more suitable for the extraction of the upper ore edge or the extraction of the lower ore edge can be obtained, the loss of the lower ore edge is reduced, the accuracy of the Canny operator in extracting the edge is improved, and the accuracy of ore detection is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a method flow chart of a powder grinding quality detection method according to an embodiment of the present application.

[0024] Figure 2 This is a method flow chart of step S1 of a powder grinding quality detection method in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The embodiment of the present application discloses a method and system for detecting the quality of powder grinding, which segments the powder image of the ore to form a plurality of blocks. After clustering the blocks, clusters are formed, and the upper layer probability of each cluster belonging to the upper layer of ore is calculated. The optimal upper threshold and the optimal lower threshold of each cluster are calculated based on the upper layer probability of the cluster, and the Canny algorithm extracts the edge based on the optimal upper threshold and the optimal lower threshold of each cluster. In this process, the optimal upper threshold and the optimal lower threshold in the Canny algorithm are automatically adjusted under the action of the upper layer probability of each cluster, thereby reducing the loss of the edge of the lower layer ore particles and improving the accuracy of ore particle detection. Reference Figure 1 The powder grinding quality detection method includes steps S1 to S3.

[0026] Step S1: collecting a powder image and segmenting the powder image to form a plurality of image blocks; The image acquisition device is used in conjunction with a telephoto macro lens to obtain a macro image of the ore particles, and the macro image of the ore particles is grayed to obtain a powder image. By graying the image, the workload of subsequent calculations is reduced and the calculation efficiency is improved.

[0027] The powder image is segmented using an image segmentation algorithm to form blocks, where the grayscale values ​​of pixels in the blocks are similar. In this embodiment, the powder image is segmented using a superpixel segmentation method. In other embodiments, the image can also be segmented using a region growing method.

[0028] Step S1: construct the grayscale feature vector of each block, perform clustering on each block, and obtain multiple classification clusters; Reference Figure 2 , step S1 includes steps S11 to S15; S11: Calculate the overall grayscale of the block; Get the grayscale value of the pixels in the tile, and take the average grayscale value of all pixels in the tile as the overall grayscale of the tile.

[0029] S12: Calculate the internal contrast of the image block; In one embodiment, the calculation formula of the internal contrast is: ; In the formula, Indicates The internal contrast of each tile; Indicates The total number of pixels in a tile; Indicates the pixel ordinal number; Indicates In the block The gray value of each pixel; Indicates The mean gray value of all pixels in a block; represents the linear normalization function.

[0030] Indicates the difference between the grayscale value of any pixel in the tile and the mean grayscale value in the tile. The difference between the grayscale value of each pixel in the tile and the mean grayscale value of the pixels in the tile is accumulated. The larger the accumulated value, the greater the difference in the grayscale values ​​of the pixels in the tile and the higher the internal contrast.

[0031] In another embodiment, the calculation formula of the internal contrast can be expressed as: ; In the formula, Indicates The internal contrast of each tile; Indicates The total number of pixels in a tile; Indicates the pixel ordinal number; Indicates In the block The gray value of each pixel; Indicates The mean gray value of all pixels in a block; represents the linear normalization function.

[0032] S13: Calculate the global deviation of the patch; The step of calculating the global deviation includes: calculating the average grayscale of all pixels in the powder image to obtain the image grayscale; and taking the difference between the overall grayscale of any block and the image grayscale as the global deviation of the block.

[0033] S14: construct feature vector; In one embodiment, the expression of the feature vector corresponding to any tile is: ; In the formula, is the overall grayscale of the block; is the internal contrast of the block; is the global deviation of the patch; In another embodiment, the expression of the feature vector corresponding to any tile is: ; In the formula, is the overall grayscale of the block; is the internal contrast of the block; is the global deviation of the patch; Represents the mean value of the gradient amplitude of the pixels in this block.

[0034] The gradient amplitude of the pixel points in the image block can be obtained through the Sobel operator. The eigenvalue of the pixel gradient is introduced into the feature vector, thereby further improving the accuracy of subsequent clustering of the feature vector.

[0035] S15: clustering the feature vectors; A clustering method is used to cluster the feature vectors to obtain multiple cluster clusters; the clustering method in this embodiment is an iterative self-organizing clustering algorithm, and in other embodiments it can also be other clustering algorithms, such as the K-means clustering method, in which the K value is set by the staff based on experience.

[0036] The feature vector of the image block contains multi-dimensional data. The image block is classified based on the multi-dimensional data of the feature vector to improve the accuracy of classification.

[0037] S2: Calculate the upper layer probability of each cluster being the upper layer ore, and calculate the upper adjustment coefficient and the lower adjustment coefficient of each cluster according to the upper layer probability of each cluster; S21: Calculate the upper layer probability; In one embodiment, the formula for calculating the upper layer probability is: ; In the formula, Indicates The upper probability of clusters; Indicates The mean of the overall grayscale of all blocks in a cluster; Indicates The mean of the global deviations of all tiles in the cluster; Indicates The mean internal contrast of all patches in a cluster; Expressed as a natural constant is the logarithmic function of the base; is a linear normalization function.

[0038] In the formula, and Multiplication is used to enhance the synergistic effect of high grayscale and high contrast. Logarithmic function Used to limit The value range of the calculation result can avoid the numerical explosion caused by excessive contrast, and make It has nonlinear growth characteristics; when And in the process of gradual increase, The value of gradually approaches 0; As it approaches 1, the probability that all tiles in the cluster belong to the upper layer of ore increases; on the contrary, hour, As it increases, the probability that all tiles in the cluster belong to the upper layer of ore decreases.

[0039] If the global deviation mean in the cluster is positive, that is, , it means that the gray value in the cluster is high, and the possibility that the cluster is the upper layer ore is greater; on the contrary, if the mean of the global deviation in the cluster is negative, that is, , it means that the gray value in the cluster is low, and the possibility that the cluster is the upper layer ore is smaller.

[0040] when hour, Approaching 1, the probability that all tiles in the cluster belong to the upper layer of ore increases significantly; when hour, As it increases, the probability that all tiles in the cluster belong to the upper layer of ore is suppressed and reduced.

[0041] In another embodiment, the upper layer probability is calculated as follows: ; In the formula, Indicates The upper probability of clusters; Indicates The mean of the overall grayscale of all blocks in a cluster; Indicates The mean of the global deviations of all tiles in the cluster; Indicates The mean internal contrast of all patches in a cluster; Expressed as a natural constant is the logarithmic function of the base; is a linear normalization function; It is an adjustment coefficient, which is used to adjust the influence of global deviation on the upper layer probability. This coefficient can be adjusted according to the actual situation to adapt to a more complex industrial production environment.

[0042] S22: calculating an upper adjustment coefficient and a lower adjustment coefficient; The calculation formula of the upper adjustment coefficient is: , where Indicates The up-regulation coefficient of each cluster; Indicates The upper probability of the clusters.

[0043] The calculation formula of the lower adjustment coefficient is: , where Indicates The down-regulation coefficient of the clusters; Indicates The upper probability of the clusters.

[0044] S3: Determine the optimal upper threshold and the optimal lower threshold of the canny operator based on the upper adjustment coefficient and the lower adjustment coefficient, and use The operator performs edge detection on the blocks in each cluster to determine whether the powder is qualified; During the edge detection process, the optimal high threshold and the optimal low threshold of each cluster are calculated according to the upper probability of each cluster, so that the optimal high threshold and the optimal low threshold can be automatically adjusted in the process of edge extraction by the Canny operator, thereby reducing the loss of the lower edge and improving the accuracy of ore particle detection.

[0045] In one embodiment, the calculation formula of the optimal high threshold is: ; express The optimal high threshold of the operator; Indicates the preset experience threshold; Indicates The up-regulation coefficient of the clusters.

[0046] The calculation formula for the optimal low threshold is: ; express The low threshold of the operator; Indicates the preset experience threshold; Indicates The down-regulation coefficient of the cluster.

[0047] In another embodiment, the calculation formula of the optimal high threshold is: ; express The optimal high threshold of the operator; Indicates the preset experience threshold; Indicates The up-regulation coefficient of the clusters.

[0048] The calculation formula for the optimal low threshold is: ; express The low threshold of the operator; represents the preset empirical threshold, and ; Indicates The down-regulation coefficient of the cluster.

[0049] In this embodiment, the staff can adjust the minimum difference between the optimal upper threshold and the optimal lower threshold according to actual conditions to improve the flexibility in the edge detection process.

[0050] During edge extraction, when the probability of a cluster being in the upper layer is high, the optimal high threshold calculated is also high, which will result in a low optimal low threshold, making the canny operator more stringent in screening edge pixels. Since the upper layer of powder is not blocked, the gradient amplitude of its edge pixels is large. Selecting a larger optimal high threshold can reduce the false detection of oversaturated areas in the powder image.

[0051] When the upper probability of a cluster is small, the optimal upper threshold decreases, the optimal lower threshold increases, and the screening conditions for edge pixels are relaxed, so that more edge pixels can be retained and the loss of lower edge layers can be reduced.

[0052] Perform a closing operation on any edge, connect the broken edges, and calculate the equivalent diameter of the edge. The calculation formula for the equivalent diameter is: ; In the formula, represents the equivalent diameter of the edge; It represents the number of all pixels contained in the edge, that is, the area of ​​the edge.

[0053] The product of the equivalent diameter of the edge and the actual length of the unit pixel is taken as the actual diameter of the edge; The actual length of the unit pixel can be obtained by using the method of calibrating the reference object, for example, obtaining a reference object with a known actual length, obtaining the pixel length of the reference object, and the ratio of the actual length of the reference object to the number of pixel blocks is the actual length of the unit pixel.

[0054] An abnormal length is set, and the edge is judged to be abnormal when the actual diameter of the edge is greater than the abnormal length; a quantity threshold is set, and when the ratio of the number of abnormal edges to the total number of edges reaches the quantity threshold, the stone is judged to be unqualified and needs further crushing and grinding.

[0055] An embodiment of the present application also discloses a powder grinding quality detection system, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a powder grinding quality detection method according to the present application is implemented.

[0056] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.

[0057] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for detecting the grinding quality of powder, characterized in that: The method comprises the following steps: collecting powder images, segmenting the powder images to form a plurality of image blocks; constructing a grayscale feature vector of each image block, performing clustering processing on each image block to obtain a plurality of classification clusters; Calculate the upper layer probability of each cluster being the upper layer ore, and determine the upper adjustment coefficient and the lower adjustment coefficient of each cluster; determine the optimal upper threshold and the optimal lower threshold of the Canny operator based on the upper adjustment coefficient and the lower adjustment coefficient, and use the Canny operator to perform edge detection on the blocks in each cluster to determine whether the powder is qualified; The calculation formula for the upper probability of the cluster is: ; In the formula, Indicates The upper probability of clusters; Indicates The mean of the overall grayscale of all blocks in a cluster; Indicates The mean of the global deviations of all tiles in the cluster; Indicates The mean internal contrast of all patches in a cluster; Indicated by natural constant is the logarithmic function of the base; is a linear normalization function.

2. A powder grinding quality detection method according to claim 1, characterized in that: The steps of calculating the overall grayscale of the image block include obtaining the grayscale value of each pixel in the image block; and taking the average of the grayscale values ​​of all the pixels in the image block as the overall grayscale of the image block.

3. A powder grinding quality detection method according to claim 2, characterized in that: The internal contrast of a tile is calculated as: ; Indicates Internal contrast in each tile; Indicates The total number of pixels in a tile; Indicates the pixel ordinal number; Indicates In the block The gray value of each pixel; Indicates The mean gray value of all pixels in a block; Represents the standard normalization function.

4. A powder grinding quality detection method according to claim 2, characterized in that: The internal contrast of a tile is calculated as: ; Indicates Internal contrast in each tile; Indicates The total number of pixels in a tile; Indicates the pixel ordinal number; Indicates In the block The gray value of each pixel; Indicates The mean gray value of all pixels in a block; Represents the standard normalization function.

5. A powder grinding quality detection method according to claim 3 or 4, characterized in that: The steps for calculating the global deviation of the image block include: obtaining the grayscale values ​​of all pixels of the powder image and calculating the average value to obtain the global grayscale; and taking the difference between the overall grayscale of the image block and the global grayscale as the global deviation.

6. A powder grinding quality detection method according to claim 5, characterized in that: The feature vector includes: the overall grayscale of the tile, the global grayscale of the tile, and the internal contrast of the tile.

7. A powder grinding quality detection method according to claim 1, characterized in that: An iterative self-organizing clustering algorithm is used to cluster the tiles using the feature vector as the metric.

8. A powder grinding quality detection method according to claim 1, characterized in that: The powder image is segmented using superpixel segmentation method.

9. A powder grinding quality detection method according to claim 1, characterized in that: In the step of calculating the upper adjustment coefficient and the lower adjustment coefficient of each cluster according to the upper probability of each cluster, the calculation formula of the upper adjustment coefficient is: , where Indicates The up-regulation coefficient of each cluster; Indicates The upper probability of clusters; The calculation formula of the lower adjustment coefficient is: , where Indicates The down-regulation coefficient of the clusters; Indicates The upper probability of the clusters.

10. A powder grinding quality detection system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a powder grinding quality detection method according to any one of claims 1 to 9 is implemented.

Images