Image-based sorted object pollution dynamic tracking method

Abstract

The invention discloses an image-based sorted object pollution dynamic tracking method, which belongs to the field of sorted object pollution monitoring, and is characterized in that a pollution identification area is arranged on a sorted object conveying line, and image acquisition, edge detection and 3D point cloud data projection technologies are combined, so that the actual contour of a sorted object is accurately identified; by extracting color, texture and shape characteristic parameters of an appearance image and internal spectral characteristic parameters and density distribution parameters, the appearance and internal pollution conditions of the sorted objects are quantitatively evaluated; the pollution diffusion risk is comprehensively evaluated by calculating the pollution intensity value, the self-expansion evaluation value and the external expansion evaluation value of the polluted sorted objects; and finally, the positions of the polluted sorted objects are tracked in real time, task processing queues of the sorting stations are sorted according to pollution evaluation values, the optimal sorting station is intelligently distributed, the sorting efficiency is improved, and the pollution diffusion risk is reduced. The problems that the sorting object contour recognition precision is insufficient, the pollution degree judgment is not specific enough, and sorting station distribution is rigid are effectively solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of sorting material pollution monitoring, and particularly relates to a sorting material pollution dynamic tracking method based on images. BACKGROUND

[0002] In the process of industrial production, logistics sorting and material processing, the sorting material pollution problem has always been a key factor affecting product quality and production efficiency. The traditional pollution detection method has the following limitations. 1. The sorting material contour recognition accuracy is insufficient and the pollution area judgment is one-sided. 2. The sorting material pollution degree judgment is not specific enough. 3. The sorting station distribution of the pollution sorting material is rigid and low in efficiency. Therefore, we propose a sorting material pollution dynamic tracking method based on images. SUMMARY

[0003] The purpose of the present application is to provide a sorting material pollution dynamic tracking method based on images to solve the problems raised in the background art.

[0004] To achieve the above purpose, the present application provides the following technical scheme: a sorting material pollution dynamic tracking method based on images, comprising the following steps: Step 1: dividing the sorting material conveying line into conveying areas and identifying the actual contour of the sorting material in each conveying area; analyzing the appearance pollution evaluation value and the internal pollution evaluation value of the sorting material, and judging whether the appearance and the interior of the sorting material are polluted accordingly; Step 2: monitoring the area and type of each pollution area of the pollution sorting material, calculating the pollution intensity value of the pollution sorting material, combining the area, perimeter and edge definition of each pollution area, calculating the self-expansion evaluation value of the pollution sorting material, combining the number of sorting materials in the conveying area where the pollution sorting material is located to calculate the external expansion evaluation value, and calculating the pollution evaluation value of the pollution sorting material accordingly; Step 3, real-time tracking the position of the pollution sorting material, distributing the sorting stations in turn according to the order of the appearance of the pollution sorting material, analyzing the time length of the pollution sorting material to reach each sorting station, the task processing efficiency of the sorting station and the queuing sequence in the sorting station, and evaluating the optimal sorting station for processing the pollution sorting material accordingly.

[0005] Preferably, in step 1, the specific process of identifying the actual contour of the sorting material in each conveying area is as follows: A pollution identification area is set on the sorting material conveying line, and the sorting material conveying line is evenly divided into multiple conveying areas; For each conveying area conveying to the pollution identification area, the image of the conveying area is collected, and the preliminary contour of the sorting material in the conveying area is outlined by using edge detection technology; The 3D point cloud data of the sorted object is acquired, and the 3D point cloud data is projected into a two-dimensional image coordinate system so as to be in the same coordinate system as the preliminary contour; the best matching relationship between the 3D point cloud data and the preliminary contour is determined, and the preliminary contour is adjusted and corrected according to the best matching relationship, so as to obtain the actual contour of the sorted object.

[0006] Preferably, in step one, the specific process of judging whether the appearance of the sorted object is contaminated is as follows: For each sorted object in the conveying area, an appearance image of the actual contour of the sorted object is collected, and color feature parameters, texture feature parameters and shape feature parameters of the appearance image are extracted; The color feature parameters, texture feature parameters and shape feature parameters of the appearance image of the sorted object are matched with the standard range of the corresponding feature parameters, the feature parameters not in the corresponding standard range are extracted, and the deviation values of the feature parameters are quantified to obtain appearance feature deviation values; After the appearance feature deviation values corresponding to each appearance feature parameter are normalized and de-dimensioned, the appearance pollution evaluation value WP is obtained by using the formula: , wherein i is the index of the appearance feature parameter, n is the total number of the appearance feature parameters, PLi is the appearance feature deviation value corresponding to the i-th appearance feature parameter, and wi is the preset weight coefficient corresponding to the i-th appearance feature parameter; If the appearance pollution evaluation value is greater than the corresponding threshold value, it is judged that the appearance of the sorted object is contaminated. The above method of judging whether the appearance of the sorted object is contaminated is recorded as a pollution evaluation method.

[0007] Preferably, in step one, the specific process of judging whether the interior of the sorted object is contaminated is as follows: For each sorted object in the conveying area, the spectral feature parameters of the interior of the sorted object and the density distribution parameters of the sorted object are collected; The standard interior spectral feature parameters and the density distribution parameter range of the sorted object are acquired; the interior pollution evaluation value of the sorted object is calculated by using the pollution evaluation method, and is compared with the corresponding threshold value to judge whether the interior of the sorted object is contaminated.

[0008] Preferably, in step two, the specific process of calculating the pollution intensity value of the contaminated sorted object is as follows: The sorted object contaminated is marked as a contaminated sorted object, and each contaminated area of the contaminated sorted object is detected to obtain the area and the type of each contaminated area of the current contaminated sorted object; A pollution type risk level table is constructed, and each pollution type in the pollution type risk level table corresponds to a risk weight value; For each pollution classification, the pollution type of each pollution area corresponding to the pollution classification is matched by being substituted into the pollution type risk level table, and the risk weight value of each pollution area is output; The pollution intensity value WQ is obtained by multiplying the area of each pollution area of the pollution classification by the risk weight value of the corresponding pollution area type and then adding them together.

[0009] Preferably, the specific process of calculating the self-expansion evaluation value and the external expansion evaluation value of the pollution separation object is as follows: For each pollution area of the pollution separation object, the perimeter of the pollution area is obtained, and the compactness of the pollution area is obtained by dividing the square of the perimeter by the area; The edge definition of the pollution area is obtained; The self-expansion evaluation value of the pollution area is obtained by multiplying the compactness and the edge definition of the pollution area by the corresponding preset self-expansion potential coefficient and then adding them together; The mean value of the self-expansion evaluation values of all pollution areas of the pollution separation object is calculated to obtain the self-expansion evaluation value ZS of the pollution separation object; The number of separation objects in the conveying area where the pollution separation object is located is counted to obtain the external expansion target value; The external expansion evaluation value WS of the pollution separation object is obtained by multiplying the external expansion target value of the pollution separation object by the corresponding preset external expansion potential coefficient.

[0010] Preferably, in step two, the calculation process of the pollution evaluation value is as follows: The pollution intensity value WQ corresponding to the pollution separation object, the self-expansion evaluation value ZS of the pollution separation object, and the external expansion evaluation value WS of the pollution separation object are normalized and de-dimensioned, and then the formula WPZ=WQ×b1+ZS×b2+WS×b3 is used to obtain the pollution evaluation value WPZ, wherein b1, b2, and b3 are preset weight coefficients.

[0011] Preferably, in step three, the specific process of analyzing the time length of the pollution separation object reaching each sorting station, the task processing efficiency of the sorting station, and the queuing sequence in the sorting station is as follows: The current position of each pollution separation object is obtained in real time, and multiple sorting stations are arranged on the separation object conveying line; For each sorting station, each pollution classification in the corresponding task processing queue is sorted according to the corresponding pollution evaluation value to obtain a task processing priority sequence; Each pollution separation object is determined in order according to the order of appearance of the pollution separation object, and the time length of the current pollution separation object moving to each sorting station is obtained, which is recorded as the expected arrival time YT; By inserting the current pollution sorting object into the task processing priority sequence of each sorting station according to the size of the pollution evaluation value corresponding to the current pollution sorting object, the queuing sequence PD of the current pollution sorting object in each sorting station is obtained. An average processing efficiency of each sorting station processing the pollution sorting object is obtained, and is denoted as task processing efficiency XL.

[0012] Preferably, the specific process of evaluating the optimal sorting station processing the pollution sorting object is as follows: For each sorting station, after the expected arrival time YT of the pollution sorting object, the task processing efficiency XL of the sorting station and the queuing sequence PD of the pollution sorting object in the current sorting station are normalized and de-dimensioned, the formula is used: , to obtain the station selection evaluation value XPZ, wherein c1, c2 and c3 are preset weight coefficients. The sorting station with the maximum station selection evaluation value is selected as the processing station of the current pollution sorting object, and the task execution instruction of the pollution sorting object is sent to the control execution end of the corresponding sorting station to trigger the sorting operation.

[0013] Compared with the prior art, the present application has the following advantages: (1) The image-based sorting object pollution dynamic tracking method can accurately identify the actual contour of the sorting object by combining image acquisition, edge detection and 3D point cloud data projection technology; further, the ICP matching algorithm is used to adjust and correct the preliminary contour, effectively improving the accuracy of contour recognition; at the same time, by extracting the color, texture and shape feature parameters of the appearance image and matching them with the standard range, the quantitative evaluation of appearance pollution is realized; in addition, by collecting the internal spectral feature parameters and density distribution parameters, it can accurately judge whether there is pollution inside the sorting object; this inside-out detection method comprehensively improves the accuracy and reliability of the sorting object pollution detection.

[0014] (2) The image-based sorting object pollution dynamic tracking method can monitor the area and type of each pollution area of the pollution sorting object, construct a pollution type risk level table, calculate the pollution intensity value, calculate the compactness by combining the perimeter and area of the pollution area, obtain the edge definition, calculate the self-expansion evaluation value of the pollution sorting object, and comprehensively evaluate the pollution diffusion risk by counting the number of sorting objects in the conveying area where the pollution sorting object is located and combining the preset external expansion potential coefficient to calculate the external expansion evaluation value; by comprehensively analyzing the pollution intensity value, the self-expansion evaluation value and the external expansion evaluation value, the pollution evaluation value is calculated, which comprehensively reflects the pollution degree and diffusion risk of the pollution sorting object, provides a scientific basis for subsequent classification decision, and effectively improves the comprehensiveness and accuracy of the pollution risk evaluation.

[0015] (3) The image-based sorting pollution dynamic tracking method sorts the polluted sorting objects in the task processing queue according to the pollution evaluation value for each sorting station, forms a task processing priority sequence, and ensures that the sorting objects with high pollution risk are processed preferentially; the expected arrival time length of the polluted sorting objects to each sorting station, the processing efficiency of the sorting station, and the queuing sequence are comprehensively considered, and a station selection evaluation value is obtained through comprehensive analysis, the time cost, processing efficiency and waiting time are fully weighed, and scientific evaluation and optimal selection of the sorting station are realized; the sorting station with the maximum station selection evaluation value is selected as the processing station, and a task execution instruction is sent in time to trigger the sorting operation, so that the efficient execution of the sorting process is ensured; this method not only improves the utilization efficiency of the sorting station, but also maximally reduces the risk of pollution diffusion, and realizes the unity of efficiency and flexibility of the sorting operation. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The flowchart of the present application. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0018] Embodiment: please refer to Figure 1 The present application provides an image-based sorting pollution dynamic tracking method, which comprises: Step 1: divide the sorting object conveying line into conveying regions, and identify the actual contour of the sorting objects in each conveying region; analyze the appearance pollution evaluation value and the internal pollution evaluation value of the sorting objects, and judge whether the appearance and the interior of the sorting objects are polluted according to the above, and the specific process is as follows: A pollution identification area is arranged on the sorting object conveying line, and the sorting object conveying line is evenly divided into a plurality of conveying regions; For each conveying region conveying to the pollution identification area, the image of the conveying region is collected, and the preliminary contour of the sorting object in the conveying region is outlined by using edge detection technology; The 3D point cloud data of the sorting object is obtained, and the 3D point cloud data is projected into a two-dimensional image coordinate system, so that it is in the same coordinate system with the preliminary contour; the best matching relationship between the 3D point cloud data and the preliminary contour is determined by using ICP matching algorithm, and the preliminary contour is adjusted and corrected according to the matching result, and the actual contour of the sorting object is obtained; Further, the process of adjusting and correcting the preliminary contour according to the matching result is as follows: The correspondence between the points in the 3D point cloud data and the points on the two-dimensional preliminary contour is determined according to the ICP matching algorithm, the preliminary contour is aligned with the 3D point cloud data by corresponding translation and rotation in the coordinate system, and the shape of the preliminary contour is adjusted according to the corresponding point set; if the 3D point cloud data shows that a part of the preliminary contour is missing or deformed, the part of the contour can be supplemented or corrected according to the position information of the corresponding points; For each sorting object in the conveying area, an appearance image of the sorting object is collected using a color camera, and color feature parameters, texture feature parameters and shape feature parameters of the appearance image are extracted; Further, the color feature parameters include brightness, chroma and saturation; the texture feature parameters include contrast, correlation, energy and entropy; and the shape feature parameters include area, perimeter, aspect ratio and circularity; The color feature parameters, texture feature parameters and shape feature parameters of the appearance image of the sorting object are matched with the standard range of the corresponding feature parameters, the feature parameters not in the corresponding standard range are extracted, and the deviation values of the feature parameters are quantified to obtain the appearance feature deviation values; After the appearance feature deviation values corresponding to each appearance feature parameter are normalized and de-dimensioned, the appearance pollution evaluation value WP is obtained by using the formula: , wherein i is the index of the appearance feature parameter, n is the total number of the appearance feature parameters, PLi is the appearance feature deviation value corresponding to the i-th appearance feature parameter, and wi is the preset weight coefficient corresponding to the i-th appearance feature parameter; A preset appearance pollution evaluation threshold is set, and if the appearance pollution evaluation value is greater than the corresponding threshold, the appearance pollution of the sorting object is determined; The above method for determining the appearance pollution of the sorting object is referred to as the pollution evaluation method; For each sorting object in the conveying area, the internal spectral feature parameters of the sorting object are collected using a near-infrared spectrometer, and the density distribution parameters of the sorting object are measured using a capacitance sensor; the standard internal spectral feature parameters and the density distribution parameter range of the sorting object are obtained; the internal pollution evaluation value of the sorting object is calculated using the pollution evaluation method, and compared with the corresponding threshold to determine whether the internal pollution of the sorting object is present; Further, the process of determining whether the internal pollution of the sorting object is present is as follows: The spectral feature parameters include absorption peak position and intensity at a specific wavelength, reflection peak position and intensity, slope of the spectral curve, and spectral reflectance; The density distribution parameters include voxel density, average density, density distribution curve, local density change rate, density contrast, and density uniformity index; The spectral feature parameters and the density distribution parameters inside the sorted objects are matched with the standard range of the corresponding internal feature parameters, the internal feature parameters not within the corresponding standard range are extracted, and the deviation value of the feature parameters is quantified to obtain the internal feature deviation value; The internal pollution evaluation value is obtained by normalizing and de-dimensioning the feature deviation values corresponding to each feature parameter and then performing weighted calculation, A preset internal pollution evaluation threshold is set, and if the internal pollution evaluation value is greater than the corresponding threshold, it is determined that the sorted object is internally polluted.

[0019] It should be noted that by setting a pollution identification area on the sorted object conveying line and combining image acquisition and edge detection technology, the actual contour of the sorted object can be accurately identified; further adjusting and correcting the preliminary contour using 3D point cloud data and ICP matching algorithm effectively improves the accuracy of contour identification, providing accurate morphological basis for subsequent pollution detection; At the same time, the appearance and internal pollution of the sorted object are analyzed, the color, texture and shape feature parameters of the appearance image are extracted, and are matched with the standard range to realize the quantitative evaluation of the appearance pollution; similarly, by collecting the spectral feature parameters and the density distribution parameters inside, it can be accurately judged whether there is pollution inside the sorted object; this inside-out detection method comprehensively improves the accuracy and reliability of the sorted object pollution monitoring.

[0020] Step two: monitor the area and type of each pollution area of the polluted sorted object, calculate the pollution intensity value of the polluted sorted object, combine the area, perimeter and edge definition of each pollution area, calculate the self-expansion evaluation value of the polluted sorted object, combine the number of sorted objects in the conveying area where the polluted sorted object is located to calculate the external expansion evaluation value, and calculate the pollution evaluation value of the polluted sorted object accordingly, the specific process is as follows: The sorted object affected by pollution is marked as a polluted sorted object, and each pollution area of the polluted sorted object is detected to obtain the area and pollution type of each pollution area of the current polluted sorted object; A pollution type risk level table is constructed, and a risk weight value corresponding to each pollutant is preset in the pollution type risk level table; For each pollution classification object, the pollution type of each pollution area corresponding thereto is matched by substituting it into the pollution type risk level table to output the risk weight value of each pollution area; The pollution intensity value WQ is obtained by multiplying the area of each pollution area of the pollution classification object by the risk weight value of the corresponding pollution area and then adding them up; For each pollution area of the polluted sorted object, the perimeter of the pollution area is obtained, and the compactness of the pollution area is obtained by dividing the square of the perimeter by the area; the greater the compactness of the pollution area, the more irregular the pollution area, and the more easily it spreads; Obtain the edge sharpness of the contaminated area; where edge sharpness reflects the clarity of the boundary between the contaminated area and the surrounding uncontaminated area. A preset self-expansion potential coefficient is assigned to the edge clarity and compactness of the contaminated area, respectively. The preset self-expansion potential coefficient can be adjusted according to the type of contamination, the material of the sorted items, the area of ​​contamination, etc. The self-diffusion evaluation value of the contaminated area is obtained by multiplying the compactness and edge clarity of the contaminated area by the corresponding preset self-expansion potential coefficient and then adding them together. Then, the average value of the self-diffusion evaluation values ​​of all contaminated areas of the contaminated sorted material is calculated to obtain the self-expansion evaluation value ZS of the contaminated sorted material. The higher the self-expansion evaluation value of the contaminated sorted material, the higher its risk of contamination spread; the compactness and edge clarity of the contaminated area were normalized and dimensionless before the calculation process. The number of sorted items within the transport area containing contaminated sorted items is counted to obtain the target value for expansion; an expansion potential coefficient is preset, which is controlled by the type of contamination of the sorted items, the direction of movement of the sorted items, and the speed of movement. The contamination spread assessment value WS is obtained by multiplying the target value of the contaminated sorted material by the corresponding spread potential coefficient; the larger the contamination spread assessment value of the contaminated sorted material, the higher the risk of contamination spread of the contaminated sorted material to other sorted materials. After normalizing and removing the dimensions of the pollution intensity value WQ, the self-expansion evaluation value ZS, and the external expansion evaluation value WS of the polluted sorted items, the pollution evaluation value WPZ is obtained using the formula: WPZ=WQ×b1+ZS×b2+WS×b3, where b1, b2, and b3 are preset weighting coefficients. The specific process for detecting polluted areas and types of pollutants using pollutant classification materials is as follows: For sorted items with visible contamination: Obtain images of the actual outlines of the visible contaminants and preprocess the acquired images. Obtain the RGB values ​​of each pixel in the appearance image and compare them with the RGB values ​​of the corresponding pixel in the corresponding standard image; if the deviation between the two is greater than the corresponding preset threshold, the pixel is determined to be a contaminated point. A distance-based clustering method is used to cluster adjacent contamination points to obtain contamination point clusters; For each group of contamination point clusters, an edge detection algorithm is used to extract the boundaries of the contamination point clusters to obtain the apparent contamination area, and the contamination area of ​​the apparent contamination area is extracted. For each piece of appearance pollution area, the pollution matching parameters of each pixel point in the appearance pollution area are extracted, and the pollution mean of each pollution matching parameter is calculated, and the obtained mean of each pollution matching parameter is used to construct a feature matching vector of the appearance pollution area; wherein the pollution matching parameters include: RGB value, brightness value, saturation and the like of the pixel point; Collect and analyze sample data of various known pollution types, extract the appearance feature matching vector template for each pollution type, combine each pollution type and its corresponding appearance feature matching vector template to construct an appearance pollution type matching library; For each piece of appearance pollution area, the corresponding appearance feature matching vector is substituted into the appearance pollution type matching library to calculate the similarity with all appearance feature matching vector templates, and the pollution type corresponding to the appearance feature matching vector template with the highest similarity value is output as the pollution type of the current appearance pollution area.

[0021] For the sorting object with internal pollution, based on the three-dimensional point cloud data of the sorting object, a three-dimensional model of the internal structure of the sorting object is constructed, and the three-dimensional region inside the sorting object is divided into a plurality of voxels; The collected spectral feature parameter data is mapped to the corresponding voxels of the three-dimensional model to ensure that each voxel contains complete spectral information; The spectral curve of each voxel is analyzed to extract the intensity values of the absorption peaks and reflection peaks; The extracted intensity values of the absorption peaks and reflection peaks are compared with the preset standard range respectively; If the intensity value of the absorption peak or the reflection peak of a voxel exceeds the corresponding standard range, the voxel is marked as an internal potential pollution voxel; Based on the distance-based clustering method, adjacent pollution voxels are clustered to obtain a pollution voxel cluster; For each group of pollution voxel clusters, an edge detection algorithm is used to extract the boundary of the pollution voxel cluster to obtain an internal pollution area, and the pollution area of the internal pollution area is extracted; For each piece of internal pollution area, an internal matching feature vector is extracted, which includes the mean of the spectral feature parameters and the mean of the density feature parameters of the internal pollution area and the like; An internal pollution type matching library is constructed, which contains all known classification objects internal pollution types, and each internal pollution type corresponds to an internal matching feature vector template; The internal feature matching vector of the internal pollution area is substituted into the internal pollution type matching library for matching, and the pollution type corresponding to the most similar internal feature matching vector template is output as the pollution type of the current internal pollution area; The sorting object with internal and external complex pollution is simultaneously subjected to internal and external pollution monitoring of the sorting object.

[0022] It should be noted that by monitoring the area and type of each pollution area of the pollution classification object, and constructing a pollution type risk level table, the pollution intensity value is calculated, which fully considers the difference in harmfulness of different pollution types, and realizes accurate assessment of the pollution degree; The compactness is calculated by combining the perimeter and area of the pollution area, and the edge definition is obtained, the self-expansion evaluation value of the pollution sorting object is calculated, and the diffusion risk of the pollution in the sorting object itself is effectively predicted; By counting the number of sorting objects in the conveying area where the pollution sorting object is located, and combining the preset external expansion potential coefficient to calculate the external expansion evaluation value, strong support is provided for comprehensive assessment of the pollution diffusion risk; By weighting and summing the pollution intensity value, the self-expansion evaluation value and the external expansion evaluation value, the pollution evaluation value is calculated; this comprehensive evaluation index comprehensively reflects the pollution degree and diffusion risk of the pollution classification object, and provides a scientific basis for subsequent classification decision-making, effectively improving the comprehensiveness and accuracy of the pollution risk assessment.

[0023] Step three: Real-time tracking of the position of the pollution sorting object, according to the order of the appearance of the pollution sorting object, the sorting station is allocated in turn, the time length of the pollution sorting object reaching each sorting station, the processing efficiency of the sorting station and the queuing sequence are analyzed, and the optimal sorting station for processing the pollution sorting object is evaluated accordingly. The specific process is as follows: Install a position sensor on the sorting object conveying line to obtain the current position of each pollution sorting object in real time; At the same time, multiple sorting stations are arranged on the sorting object conveying line; Among them, the sorting station is a station that can sort out the pollution sorting object and perform pollution treatment; For each sorting station, sort the pollution classification objects in the task processing queue according to the corresponding pollution evaluation value to obtain the task processing priority sequence; According to the order of the appearance of the pollution sorting object, the sorting station is determined for each pollution sorting object in turn; For each pollution sorting object determined by the sorting station, the time length of the current pollution sorting object moving to each sorting station is obtained, which is denoted as the predicted arrival time YT; By inserting the current pollution sorting object into the task processing priority sequence of each sorting station according to the size of its corresponding pollution evaluation value, the queuing sequence PD of the current pollution sorting object in each sorting station is obtained; Get the average processing efficiency of each sorting station for processing the pollution sorting object, denoted as the task processing efficiency XL; For each sorting station, the predicted arrival time YT of the pollution sorting object, the task processing efficiency XL of the sorting station, and the queuing sequence PD of the pollution sorting object in the current sorting station are normalized and de-dimensioned, and the formula: , to obtain a workstation selection evaluation value XPZ, wherein c1, c2, and c3 are preset weight coefficients; The maximum staining workstation is selected as the processing workstation of the contaminated sorting object, and a task execution instruction of the contaminated sorting object is sent to a control execution end of the corresponding staining workstation to trigger the staining operation.

[0024] It should be noted that for each sorting workstation, the contaminated sorting objects in the task processing queue are sorted according to the pollution evaluation value to form a task processing priority sequence. This strategy ensures that sorting objects with high pollution risk can be processed preferentially, effectively improving the pertinence and timeliness of the sorting operation. The expected arrival time of the contaminated sorting object to each staining workstation, the processing efficiency of the staining workstation, and the queuing sequence are comprehensively considered to obtain the workstation selection evaluation value through normalization processing and weighted calculation. This process fully considers the time cost, processing efficiency, and waiting time, and realizes scientific evaluation and optimal selection of the staining workstation. By selecting the staining workstation with the maximum workstation selection evaluation value as the processing workstation and sending the task execution instruction in time to trigger the staining operation, efficient execution of the staining process is ensured. This method not only improves the utilization efficiency of the staining workstation, but also maximally reduces the risk of pollution diffusion, realizing the unity of efficiency and flexibility of the staining operation.

[0025] Although embodiments of the present application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An image-based dynamic tracking method of contamination of items for sorting, characterized in that: The method comprises the following steps: Step 1: Dividing the conveying area of the sorting conveying line, identifying the actual profile of the sorting material in each conveying area; analyzing the appearance pollution evaluation value and the internal pollution evaluation value of the sorting material, and judging whether the appearance and the internal of the sorting material are polluted or not; Step 2: Monitoring the area and type of each pollution area of the polluted sorting material, calculating the pollution intensity value of the polluted sorting material, combining the area, perimeter and edge definition of each pollution area to calculate the self-expansion evaluation value of the polluted sorting material, combining the quantity of the sorting material in the conveying area where the polluted sorting material is located to calculate the external expansion evaluation value, and calculating the pollution evaluation value of the polluted sorting material; Step 3: Real-time tracking the position of the polluted sorting material, distributing the sorting stations in turn according to the order of the appearance of the polluted sorting material, analyzing the time length of the polluted sorting material to reach each sorting station, the task processing efficiency of the sorting station and the queuing sequence in the sorting station, and evaluating the optimal sorting station for processing the polluted sorting material.

2. The image-based dynamic tracking of contamination of items for sorting method of claim 1, wherein: In step 1, the specific process of identifying the actual profile of the sorting material in each conveying area is as follows: A pollution identification area is arranged on the sorting conveying line, and the sorting conveying line is evenly divided into multiple conveying areas; For each conveying area conveying to the pollution identification area, the image of the conveying area is collected, and the preliminary profile of the sorting material in the conveying area is outlined by using edge detection technology; 3D point cloud data of the sorting material is obtained, the 3D point cloud data is projected into a two-dimensional image coordinate system, so that it is in the same coordinate system as the preliminary profile; the best matching relationship between the 3D point cloud data and the preliminary profile is determined, and the preliminary profile is adjusted and corrected according to the best matching relationship to obtain the actual profile of the sorting material.

3. The image-based dynamic tracking of contamination of items for sorting method of claim 2, wherein: In step 1, the specific process of judging whether the appearance of the sorting material is polluted or not is as follows: For each sorting material in the conveying area, the appearance image of the actual profile of the sorting material is collected, and the color feature parameters, texture feature parameters and shape feature parameters of the appearance image are extracted; The color feature parameters, texture feature parameters and shape feature parameters of the appearance image of the sorting material are matched with the standard range of the corresponding feature parameters, the feature parameters not in the corresponding standard range are extracted, the deviation value of the feature parameters is quantized, and the appearance feature deviation value is obtained; By normalizing and de-dimensioning the appearance feature deviation values corresponding to each appearance feature parameter, an appearance pollution evaluation value WP is obtained by using the formula: , wherein i is the index of the appearance feature parameter, n is the total number of the appearance feature parameters, PLi is the appearance feature deviation value corresponding to the i-th appearance feature parameter, and wi is the preset weight coefficient corresponding to the i-th appearance feature parameter. If the appearance pollution evaluation value is greater than the corresponding threshold value, it is judged that the appearance of the sorting material is polluted. The above method for judging whether the appearance of the sorting material is polluted or not is recorded as the pollution evaluation method.

4. The image-based dynamic tracking of contamination of items for sorting method of claim 3, wherein: In step 1, the specific process of judging whether the internal of the sorting material is polluted or not is as follows: For each sorting material in the conveying area, the spectral feature parameters of the internal of the sorting material are collected, and the density distribution parameters of the sorting material are measured; The standard internal spectral feature parameters and the density distribution parameter range of the sorting material are obtained; the internal pollution evaluation value of the sorting material is calculated by using the pollution evaluation method, and is compared with the corresponding threshold value to judge whether the internal of the sorting material is polluted or not.

5. The image-based dynamic tracking of contamination of items sorting method according to claim 4, characterized in that: In step 2, the specific process of calculating the pollution intensity value of the polluted sorting material is as follows: The sorting material subjected to pollution is marked as a polluted sorting material, each pollution area of each polluted sorting material is detected, and the area and pollution type of each pollution area of the current polluted sorting material are obtained; A pollution type risk level table is constructed, and each pollution type in the pollution type risk level table corresponds to a risk weight value; For each pollution type, the pollution type risk level table is matched by substituting the pollution types of each pollution area corresponding to the pollution type, and the risk weight values of each pollution area are outputted; The pollution intensity value WQ is obtained by multiplying the area of each pollution area of the pollution type by the risk weight value of the corresponding pollution area type and then adding them up.

6. The image-based dynamic tracking of contamination of items sorting method according to claim 5, characterized in that: The specific process of calculating the self-expansion evaluation value and the external expansion evaluation value of the pollution sorting object is as follows: For each pollution area of the pollution separation object, the perimeter of the pollution area is obtained, and the compactness of the pollution area is obtained by dividing the square of the perimeter by the area; The edge definition of the pollution area is obtained; The self-expansion evaluation value of the pollution area is obtained by multiplying the compactness and the edge definition of the pollution area by the corresponding preset self-expansion potential coefficient and then adding them up; The self-expansion evaluation value ZS of the pollution sorting object is obtained by averaging the self-expansion evaluation values of all pollution areas of the pollution sorting object. The number of sorting objects in the conveying area where the pollution sorting object is located is counted to obtain the external expansion target value. The external expansion evaluation value WS of the pollution sorting object is obtained by multiplying the external expansion target value of the pollution sorting object by the corresponding preset external expansion potential coefficient.

7. The image-based dynamic tracking of contamination of items sorting method according to claim 6, characterized in that: In step two, the calculation process of the pollution evaluation value is as follows: After the pollution intensity value WQ, the self-expansion evaluation value ZS of the pollution sorting object, and the external expansion evaluation value WS of the pollution sorting object are normalized and de-dimensioned, the pollution evaluation value WPZ is obtained by using the formula: WPZ=WQ×b1+ZS×b2+WS×b3, wherein b1, b2, and b3 are preset weight coefficients.

8. The image-based dynamic tracking of contamination of items sorting method according to claim 7, characterized in that: In step three, the specific process of analyzing the time length of the pollution sorting object reaching each sorting station, the task processing efficiency of the sorting station, and the queuing sequence in the sorting station is as follows: The current position of each pollution sorting object is obtained in real time, and multiple sorting stations are arranged on the sorting object conveying line; For each sorting station, each pollution type in the task processing queue corresponding to the sorting station is sorted according to the corresponding pollution evaluation value to obtain a task processing priority sequence; Each pollution sorting object is determined in turn according to the order of appearance of the pollution sorting object, and the time length of the current pollution sorting object moving to each sorting station is obtained, which is denoted as the expected arrival time length YT; The current pollution sorting object is inserted into the task processing priority sequence of each sorting station according to the size of the corresponding pollution evaluation value to obtain the queuing sequence PD of the current pollution sorting object in each sorting station; The average processing efficiency of each sorting station for processing pollution sorting objects is obtained, which is denoted as the task processing efficiency XL.

9. The image-based dynamic tracking of contamination of items sorting method according to claim 8, characterized in that: The specific process of evaluating the optimal sorting station for processing the pollution sorting object is as follows: For each sub-practice station, the expected arrival time YT corresponding to the contaminated sorting material, the task processing efficiency XL of the sub-practice station, and the queuing sequence PD of the contaminated sorting material in the current sub-practice station are normalized and de-dimensioned, and then the formula: is used to obtain the station selection evaluation value XPZ, wherein c1, c2, and c3 are preset weight coefficients. The sorting station with the maximum station evaluation value is selected as the processing station of the current pollution sorting object, and the task execution instruction of the pollution sorting object is sent to the control execution end of the corresponding sorting station to trigger the sorting operation.

Images