Real-time image analysis system for mineral processing

Abstract

The application discloses a kind of mineral processing real-time image analysis system, by using MeanShift algorithm filtering to obtain smooth image, by color analysis, the area of each target block degree in target label image is obtained, area, the average value of original image HSV, histogram and other characteristics, by DBSCAN density clustering analysis to the characteristics, the clustering center is obtained, the error distance of each block and clustering center is analyzed, removes noise block, obtains the outline, area and other parameters of final block, improves the rock block degree detection problem in image processing process, can better solve the phenomenon of rock block degree gathering, can realize rock block degree real-time measurement, save manpower cost, significantly improve economic benefit.

Description

Technical Field

[0001] This invention relates to the field of image processing, and more particularly to a real-time image analysis system for mineral processing. Background Technology

[0002] With the continuous development of automation and intelligent technologies, mineral processing equipment is also constantly being updated and iterated. In traditional mineral processing, the size of the ore particles is not marked during the process, which basically meets the technological requirements of traditional mineral processing equipment. However, as mineral processing equipment continues to develop with the trend of digitalization and intelligence, higher requirements are placed on the mineral processing process, necessitating the marking of ore particle size at the source, so as to trace the ore from mining to refining. Therefore, it is necessary to statistically quantify the volume of ore during the mineral processing process to improve the quality of mineral processing.

[0003] Existing methods for ore size statistics are generally achieved through machine vision. This involves acquiring images of mineral processing, identifying the ore in the images, and then calculating the size of the ore blocks based on the identification results. While this method can reduce labor costs, its accuracy is easily affected by image noise. In other words, once the acquired images are distorted, the accuracy of machine vision in calculating ore size will be greatly reduced.

[0004] Therefore, the susceptibility of existing machine vision statistical ore size methods to image noise and their low accuracy have become technical problems that urgently need to be solved by those skilled in the art. Summary of the Invention

[0005] This invention provides a real-time image analysis system for mineral processing, which solves the technical problems of existing machine vision statistical ore size methods being susceptible to image noise and having low accuracy.

[0006] To solve the above-mentioned technical problems, the technical solution proposed by this invention is as follows:

[0007] A real-time image analysis system for mineral processing includes an image acquisition module and a data processing module that communicates with the image acquisition module. The image acquisition module is used to acquire images of ore during the transportation process of the ore in the ore beneficiation production line and send the ore images to the data processing module; The data processing module is used to receive and perform MeanShift smoothing filtering on the ore image, and traverse all pixels in the ore image, marking connected regions with the same color range in the ore image as the same block, obtaining multiple connected region blocks with different color ranges; calculating the rock block size distribution, color average value, and histogram of each connected region block; performing DBSCAN density clustering analysis on the rock block size distribution, color distribution, and histogram of multiple connected region blocks to obtain cluster centers; analyzing the distance error between each connected region block and the cluster center; removing noise blocks; and obtaining the outline and / or area of ​​the final block.

[0008] Preferably, the data processing module includes: Image storage unit, used to store ore images sent by the image acquisition module; The image processing unit is used to perform MeanShift smoothing filtering on the ore image stored in the image storage unit, and traverse all pixels in the ore image, mark connected regions with the same color range in the ore image as the same block, obtain multiple connected region blocks with different color ranges, assign an ID to each connected region block, generate a mask image of the corresponding connected region block after assigning the ID, and take the IDs of all connected region blocks in the mask image, and obtain the contour of the connected region block corresponding to each ID through contour lookup. The image analysis unit is used to calculate the average area and color of each connected region block corresponding to each ID based on the outline of the found connected region block; calculate the rock block size of each connected region block based on the area of ​​each connected region block corresponding to each ID; calculate the histogram of each connected region block based on the average color of each connected region block corresponding to each ID, and obtain the color distribution of each connected region block corresponding to each ID; obtain the cluster centers by performing DBSCAN density clustering analysis on the rock block size distribution, color distribution, and histogram matrix of multiple connected region blocks; analyze the error distance between each connected region block and the cluster center, remove noise blocks, and obtain the outline and area of ​​the final block.

[0009] Preferably, the ore image stored in the image storage unit is in color RGB format. Before the image processing unit performs MeanShift smoothing filtering, it first performs edge cutting on the color RGB format ore image based on the RGB values ​​of the image, and then converts the cut ore image into a grayscale image; and performs bilateral filtering on the grayscale image; and then performs MeanShift smoothing filtering on the grayscale image after bilateral filtering.

[0010] Preferably, the MeanShift smoothing filter applied to the bilaterally filtered grayscale image refers to the image processing unit performing MeanShift smoothing filter and traversal on the bilaterally filtered grayscale image in the HSV color space.

[0011] Preferably, the image analysis unit is further configured to compare the calculated final block outline and area with a preset threshold, determine the status of the mineral processing line based on the comparison result, and send the status of the mineral processing line to a control module, wherein the control module is configured to control the speed of the mineral processing line motor based on the status of the mineral processing line.

[0012] Preferably, it also includes a result output unit, a display module, and a cloud server; the image analysis unit is used to send the obtained contour and area of ​​the final block to the result output unit; the result output unit is used to perform statistics on the ore block size on the ore beneficiation production line according to the contour and area of ​​the final block, and send the statistical results to the display module for display, and also send the statistical results to the cloud server for storage.

[0013] Preferably, the data processing module and the result output unit are both located in the embedded module, and the data processing module uses edge computing analysis to process and analyze the image.

[0014] Preferably, the image acquisition module includes a CCD industrial camera, an industrial router, and an Ethernet network; the CCD industrial camera establishes communication with the industrial router, and the industrial router establishes communication with the embedded module via the Ethernet network; the CCD industrial camera is installed at the ore beneficiation production line to acquire ore images during the transportation process of the ore in the ore beneficiation production line, and sends the ore images to the data processing module of the embedded module via the industrial router and the Ethernet network.

[0015] Preferably, it also includes a light source control unit, which establishes communication with the embedded module to receive supplementary lighting control commands from the embedded module and supplementary lighting to the CCD industrial camera according to the supplementary lighting control commands.

[0016] The present invention has the following beneficial effects: 1. The real-time image analysis system for mineral processing in this invention extracts features such as block size and area, the average HSV value of the original image, and histograms using the MeanShift algorithm. It then performs DBSCAN density clustering analysis on these features to obtain cluster centers, analyzes the error distance between each block and its cluster center, and removes noise blocks. Compared to existing technologies, this invention effectively eliminates noise blocks in ore images and then statistically analyzes the block size from the noise-removed ore images, significantly improving the accuracy of ore block size statistics. 2. In the preferred embodiment, compared with the traditional cloud computing platform, the present invention directly adopts an embedded edge computing analysis method to process and analyze images, thereby improving the processing efficiency of the production line; 3. In the preferred embodiment, the present invention performs statistical analysis on the size of mineral blocks, and allows for real-time querying of mineral block size; 4. In a preferred embodiment, the embedded module in this invention can be connected to the production line PLC control unit, enabling real-time control of the production line.

[0017] In addition to the objectives, features, and advantages described above, the present invention has other objectives, features, and advantages. The invention will now be described in further detail with reference to the accompanying drawings. Attached Figure Description

[0018] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a simplified structural diagram of the embedded mineral processing real-time image analysis system provided in the preferred embodiment of the present invention.

[0019] Figure 2 This is a flowchart of the embedded mineral processing real-time image analysis system provided in the preferred embodiment of the present invention. Detailed Implementation

[0020] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.

[0021] Example 1: This embodiment discloses a real-time image analysis system for mineral processing, including an image acquisition module and a data processing module that communicates with the image acquisition module; The image acquisition module is used to acquire images of ore during the transportation process of the ore in the ore beneficiation production line and send the ore images to the data processing module; The data processing module is used to receive and perform MeanShift smoothing filtering on the ore image, and traverse all pixels in the ore image to mark connected regions with the same color range as the same block, thereby obtaining multiple connected region blocks with different color ranges; and calculate the rock block size distribution, average color value, and histogram of each connected region block. By performing DBSCAN density clustering analysis on the rock block size distribution, average color value, and histogram of multiple connected region blocks, cluster centers are obtained, the error distance between each block and the cluster center is analyzed, noise blocks are removed, and the final block outline and / or area is obtained.

[0022] The real-time image analysis system for mineral processing in this invention extracts features such as block size and area, the average HSV value of the original image, and histograms using the MeanShift algorithm. It then performs DBSCAN density clustering analysis on these features to obtain cluster centers, analyzes the error distance between each block and its cluster center, and removes noise blocks. Compared to existing technologies, this invention effectively eliminates noise blocks in ore images, and then statistically analyzes the ore block size from the noise-block-removed ore image, significantly improving the accuracy of ore block size statistics.

[0023] Example 2: Example 2 is a preferred embodiment of Example 1, and its differences from Example 1 are described in the specific structure of the real-time image analysis system for mineral processing: like Figure 1 As shown, this embodiment discloses an ARM embedded real-time image analysis system for mineral processing. The technical solution adopted is as follows: the embedded real-time image analysis device for mineral processing includes an image acquisition module, an embedded module, a production line module, a control module, a cloud platform, a display module, and a light source control unit; The production line module is a mineral processing production line, mainly consisting of a mineral transport conveyor belt; the image acquisition module mainly includes a CCD area array camera, an industrial router, and a 1000M Ethernet, which is mainly responsible for real-time image acquisition and transmission of the detection area on the production line module. The CCD industrial camera takes pictures of the detection area of ​​the production line and transmits the digital image signal to the image storage unit of the embedded module via the industrial router through the 1000M Ethernet. The embedded module is connected to the image acquisition module and mainly includes an image storage unit, an image processing unit, an image analysis unit, and a result output unit. Its main functions are to store, process, and analyze the acquired images, and output the results to the cloud and display module. The image storage unit stores the digital signals transmitted by the image acquisition module. The image processing unit queries the image digital signals from the image storage unit, performs edge segmentation, grayscale processing, pixel conversion, and other operations on the image, and transmits the data to the image analysis unit. The image analysis unit performs pixel calculations on the converted grayscale image and analyzes it against a pre-stored block size-pixel conversion formula to determine the image block size, which is then output to the result output unit. Simultaneously, it analyzes the line state based on a pre-stored image block size threshold and outputs the results to the control module. The result output unit performs counting processing based on pre-stored statistical items and outputs the results to the cloud and display module. The display module mainly includes an HDMI port and a local display unit, which is mainly used to display statistical results. The cloud mainly includes a 5G communication unit and a cloud server, which is mainly used to store historical data of statistical results. The control module mainly includes RS485 and PLC control unit. It mainly analyzes the ore block size based on the image, and then infers the production line status. The PLC control unit controls the speed of the production line motor to realize the production line feed control. The light source control unit mainly controls the light source of the CCD area array camera; As a preferred embodiment of the above scheme, the image processing unit is an ARM chip storage unit; the edge segmentation feature quantity used by the image analysis unit is the image RGB value; and the grayscale processing function used by the image analysis unit is a built-in Python processing function.

[0024] like Figure 2 As shown, the workflow of the ARM embedded mineral processing real-time image analysis system in this embodiment includes: S1 Image Acquisition: The CCD area array camera takes high-speed pictures of the detection area and transmits the digital signal of the image acquired by the CMOS to the embedded module through the router and Ethernet interface; S2 Image Storage; The embedded system's built-in image storage unit stores the transmitted digital signals into the embedded module; S3 Edge Cutting: The image analysis unit of the embedded system has a built-in image edge cutting algorithm that performs edge cutting on the captured image based on the RGB values ​​of the image to obtain the boundary of the mineral image; S4. Grayscale processing: The image analysis unit of the embedded system has a built-in image grayscale processing algorithm to convert the cut color RGB mineral image to grayscale, turning it into black. S5, Image Analysis; The embedded system's image analysis unit has a built-in image analysis algorithm to perform pixel calculations on the converted image and output the mineral block size value through the pixel-block size formula; the mineral block size value is judged to determine whether it exceeds the block size threshold. If it exceeds the threshold, it jumps to S6; if it does not exceed the threshold, it jumps to S7. Specifically, S5 includes: S51. Preprocess the image after grayscale conversion by using bilateral filtering to remove noise, enhance the edge lines of the ore image, and preserve image details as much as possible.

[0025] S52. The enhanced image is then subjected to MeanShift smoothing filtering in the HSV color space, which can neutralize colors with similar color distributions, smooth color details, and erode smaller color areas.

[0026] S53. Analyze the MeanShift smoothed image in the HSV color space, traverse all pixels of the image, and mark each block in the connected regions of the same color range with an ID, and mask the image.

[0027] S54. Take the IDs of all blocks in the mask image, obtain the contour of each ID through contour lookup, and calculate the area, average HSV of the original image, histogram and other features of each ID.

[0028] S56. Statistically analyze the area of ​​the mask image blocks to obtain the rock block size distribution, analyze the color distribution of HSV, and statistically analyze the histogram of each block.

[0029] S57. By performing cluster analysis on block size area, color distribution, and histogram DBSCAN density, the cluster centers are obtained. The error distance between each block and the cluster center is analyzed, noise blocks are removed, and the final block's outline, area, and other parameters are obtained.

[0030] S6. Linear speed control: Controls the mineral feed speed of the production line by connecting an external PLC control module through an embedded interface.

[0031] Specifically, the mineral feed rate (i.e., motor speed) of the mineral processing line is controlled according to the status of the line. The status of the mineral processing line includes unsaturated, saturated, and normal, which is determined by the mineral size value. The specific control of the ore feed rate (i.e., motor speed) based on the status of the ore processing line is as follows: Determine the required ore particle size and proportion for the production process, and statistically analyze the current ore particle size and proportion in the beneficiation line; determine whether the current ore particle size and proportion in the beneficiation line meet the required particle size and proportion; if the current ore particle size and proportion in the beneficiation line do not meet the required particle size and proportion, adjust the mineral feed rate of the production line to make the current ore particle size and proportion meet the required particle size and proportion.

[0032] Examples are given below: Assuming the production process requires ore particles of 40mm-80mm, with a proportion of no less than 70%, if the current image shows the following ore particle size distribution: 10% below 40mm, 50% between 40mm-80mm, and 40% above 80mm, it indicates incomplete crushing in the upstream process, resulting in a high proportion of large ore (which will re-enter the crushing cycle), causing the subsequent production line to operate at undercapacity. Therefore, it is necessary to send acceleration commands to the PLC via the DP bus and monitor the speed feedback signals of the production line motors in real time. Using PID and other control algorithms, stable and rapid motor adjustment can be achieved, thereby accelerating the crushing speed of the upstream crusher and the ore supply to the downstream processes to meet the normal load conditions of the production line, thus significantly improving production line efficiency.

[0033] S7. Result Statistics: Based on the pre-stored statistical range, perform statistical counting on the block-level results of the embedded output and transmit the corresponding results to the cloud.

[0034] The block size threshold is a threshold that is adjusted in real time on the production line and can be input through a human-machine interface; In summary, this invention introduces the MeanShift algorithm, uses MeanShift filtering to obtain a smooth image, and obtains the area of ​​each target block in the target labeled image through color analysis. Features such as area, average HSV of the original image, and histogram are obtained. By performing DBSCAN density clustering analysis on the features, cluster centers are obtained. The error distance between each block and the cluster center is analyzed, noise blocks are removed, and the contour, area, and other parameters of the final block are obtained. This invention improves the problem of rock block size detection in the image processing process, better solves the phenomenon of rock block size clustering, enables real-time measurement of rock block size, saves labor costs, and significantly improves economic efficiency.

[0035] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A real-time image analysis system for mineral processing, characterized in that, It includes an image acquisition module and a data processing module that establishes communication with the image acquisition module; The image acquisition module is used to acquire images of ore during the transportation process of the ore in the ore beneficiation production line and send the ore images to the data processing module; The data processing module is used to receive and perform MeanShift smoothing filtering on the ore image, and traverse all pixels in the ore image to mark connected regions with the same color range as the same block, thereby obtaining multiple connected region blocks with different color ranges; calculate the rock block size distribution, average color value, and histogram of each connected region block; perform DBSCAN density clustering analysis on the rock block size distribution, color distribution, and histogram of multiple connected region blocks to obtain cluster centers; analyze the distance error between each connected region block and the cluster center; remove noise blocks; and obtain the outline and / or area of ​​the final block. The data processing module includes: An image storage unit is used to store ore images sent by the image acquisition module; The image processing unit is used to perform MeanShift smoothing filtering on the ore image stored in the image storage unit, and traverse all pixels in the ore image, mark connected regions with the same color range in the ore image as the same block, obtain multiple connected region blocks with different color ranges, assign an ID to each connected region block, generate a mask image of the corresponding connected region block after assigning the ID, and take the IDs of all connected region blocks in the mask image, and obtain the contour of the connected region block corresponding to each ID through contour lookup. The image analysis unit is used to calculate the average area and color of each connected region block corresponding to each ID based on the outline of the found connected region block; calculate the rock block size of each connected region block based on the area of ​​each connected region block corresponding to each ID; calculate the histogram of each connected region block based on the average color of each connected region block corresponding to each ID, and obtain the color distribution of each connected region block corresponding to each ID; obtain the cluster centers by performing DBSCAN density clustering analysis on the rock block size distribution, color distribution, and histogram matrix of multiple connected region blocks; analyze the error distance between each connected region block and the cluster center, remove noise blocks, and obtain the outline and area of ​​the final block.

2. The real-time image analysis system for mineral processing according to claim 1, characterized in that, The image storage unit stores ore images in color RGB format. Before performing MeanShift smoothing filtering, the image processing unit first performs edge cutting on the color RGB format ore image based on the RGB values ​​of the image, then converts the cut ore image into a grayscale image; performs bilateral filtering on the grayscale image; and then performs MeanShift smoothing filtering on the grayscale image after bilateral filtering.

3. The real-time image analysis system for mineral processing according to claim 2, characterized in that, The MeanShift smoothing filter applied to the bilaterally filtered grayscale image refers to the image processing unit performing MeanShift smoothing filter and traversal on the bilaterally filtered grayscale image in the HSV color space.

4. The real-time image analysis system for mineral processing according to claim 3, characterized in that, The image analysis unit is also used to compare the calculated final block outline and area with a preset threshold, and determine the status of the mineral processing line based on the comparison result, and send the status of the mineral processing line to a control module. The control module is used to control the speed of the mineral processing line motor according to the status of the mineral processing line.

5. The real-time image analysis system for mineral processing according to claim 4, characterized in that, It also includes a result output unit, a display module, and a cloud server; the image analysis unit is used to send the obtained contour and area of ​​the final block to the result output unit; the result output unit is used to perform statistics on the ore block size on the ore beneficiation production line according to the contour and area of ​​the final block, and send the statistical results to the display module for display, and also send the statistical results to the cloud server for storage.

6. The real-time image analysis system for mineral processing according to claim 5, characterized in that, The data processing module and the result output unit are both located in the embedded module, and the data processing module uses edge computing analysis to process and analyze the image.

7. The real-time image analysis system for mineral processing according to claim 6, characterized in that, The image acquisition module includes a CCD industrial camera, an industrial router, and an Ethernet network. The CCD industrial camera establishes communication with the industrial router, and the industrial router establishes communication with the embedded module via the Ethernet network. The CCD industrial camera is installed at the ore beneficiation production line to acquire images of the ore during its transportation process and to send the images to the data processing module of the embedded module via the industrial router and the Ethernet network.

8. The real-time image analysis system for mineral processing according to claim 7, characterized in that, It also includes a light source control unit, which establishes communication with the embedded module to receive supplementary lighting control commands from the embedded module and supplement the CCD industrial camera with supplementary lighting according to the supplementary lighting control commands.

Images