A visual detection method based on foreign matter screening of a color sorter

Abstract

The application discloses a kind of vision detection methods based on foreign matter screening of color sorter, the image data obtained by preprocessing, material conveying three-dimensional model is established using image data to determine the accumulation of material, and the image data is analyzed and identified to determine foreign matter characteristics, and the application relates to the field of image processing.The vision detection method based on foreign matter screening of color sorter, by preprocessing the image data obtained, material conveying three-dimensional model is established using image data to determine the accumulation of material, and the image data is analyzed and identified to determine foreign matter characteristics, and it is judged whether foreign matter characteristics is located in screening material, and then according to the accumulation and foreign matter characteristic situation generation processing strategy is transmitted, can more comprehensively obtain the image data of screening material, and guarantee the efficiency and accuracy of vision detection, better realize the vision detection operation of foreign matter screening.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, specifically to a visual inspection method based on foreign object sieving in a color sorter. Background Technology

[0002] Color sorters are now widely used in various material sorting industries, successfully replacing the original manual sorting. The machine's recognition is superior to human eye judgment in terms of speed and accuracy. Based on the differences in the optical properties of materials, the color sorter automatically sorts out the discolored particles in the granular materials through photoelectric technology.

[0003] The reference patent is titled (Patent Publication No.: CN115330806A, Patent Publication Date: 2022-11-11): "A Method for Detecting Foreign Object Defects in LED Backlight Based on Computer Vision." This method obtains the boundary values ​​of grayscale in the surface image of an LED backlight screen, divides the grayscale value sequence of pixels in the surface image into continuous and discontinuous paths, obtains the first-order difference sequence of the discontinuous path based on the correlation of pixels in the discontinuous path, and obtains the encoding method of the discontinuous path and its first-order difference sequence based on the grayscale difference between adjacent pixels in the discontinuous path, the number of consecutive identical grayscale differences, the grayscale value corresponding to the first consecutive identical grayscale difference, and the grayscale value corresponding to the first different difference after consecutive identical differences. The method compresses the LED backlight screen image at multiple scales, obtains the probability of defects in the LED backlight screen based on the encoding length of each line in the image, and determines whether defects exist on the LED surface based on the probability. This method is accurate and efficient.

[0004] Based on the description in the above documents, existing color sorters generally separate materials and detection equipment through glass. When screening materials, dust on the materials will adhere to the outer surface of the glass, affecting the color sorting effect of the color sorter. Furthermore, if the thickness of the material to be screened is too high during detection, the lower layer of material will be blocked and cannot be identified, affecting the accuracy of visual recognition. Therefore, this invention provides a visual detection method based on foreign object screening in a color sorter. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a visual inspection method based on foreign object screening in color sorters. This method solves the problems of existing color sorters, which typically separate materials and inspection equipment through glass. During material screening, dust on the materials adheres to the outer surface of the glass, affecting the color sorting effect of the color sorter. Furthermore, during inspection, if the material to be screened is piled up too thickly, the lower layer of material will be obscured and cannot be identified, thus affecting the accuracy of visual recognition.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a visual inspection method based on foreign object sieving in a color sorter, specifically comprising the following steps:

[0007] A1. The material to be screened is conveyed and image information of the conveyed material is collected. After the material to be screened enters the target area, the material to be screened is identified by a color sorter to obtain multiple image data containing the material to be screened. The obtained image data is transmitted and stored in the database.

[0008] A2. Preprocess the acquired image data, use the image data to build a three-dimensional model of material conveying to determine the material accumulation situation, analyze and identify the foreign object characteristics of the image data, and determine whether the foreign object characteristics are located in the screening material. Then, generate a processing strategy based on the accumulation situation and foreign object characteristics for transmission.

[0009] A3. Generate control commands based on the processing strategy and transmit them to the equipment units used for processing to complete the material processing operation.

[0010] Preferably, the image acquisition operation for the conveyed material in A1 is as follows:

[0011] a11. Before entering the color sorter for identification, high-definition cameras are installed directly above and to the side of the conveyor table in the conveying stage to collect images;

[0012] a13. After entering the color sorter, a glass baffle is set between the color sorter and the material to be screened for separation, and the color sorter identifies the material to be screened below the glass baffle.

[0013] Preferably, the preprocessing operation for the acquired image data in A2 is as follows:

[0014] a21. Set up multi-level nodes and perform image content classification by matching the content of multi-level nodes with the parameter content of image data;

[0015] a22. Multi-level nodes are divided into first-level nodes, second-level nodes, and third-level nodes. The content of the first-level node is the image acquisition device information, the content of the second-level node is the orientation parameter information under the first-level node, and the content of the third-level node is the acquisition timestamp information under the second-level node.

[0016] a23. Match the parameter information associated with the collected image data with the content of the multi-level nodes. When the match is consistent, fill the image data into the corresponding node.

[0017] Preferably, the operation in A2 of establishing a three-dimensional model of material conveying using image data to determine the material accumulation is as follows:

[0018] B1. Establish a three-dimensional coordinate system. Set the X-axis and Y-axis at the plane where the conveyor is located. Take the point where a corner point of the image collected from the top of the image intersects with the boundary of the conveyor as the starting point. Establish the X-axis at the point where the starting point is parallel to the boundary of the conveyor. Establish the Y-axis perpendicular to the X-axis, intersecting at the starting point and extending to the opposite boundary of the conveyor. Establish the Z-axis vertically upward from the starting point.

[0019] B2. Extract image data collected at the same time stamp, analyze and determine the feature parts, and then import the analyzed image data into the three-dimensional coordinate system according to the corresponding orientation to construct a three-dimensional model of material conveying.

[0020] B3. Extract the parameter thresholds of the current material from the historical data in the database, and compare the material characteristic parameters in the 3D model with the parameter thresholds of the material to determine the stacking situation.

[0021] Preferably, the operation in B2 to analyze the image data and determine the feature parts is as follows:

[0022] b21. Extract image data at the same timestamp, perform grayscale processing on the image data, and set multiple reference line segments parallel to the vertical boundary at equal intervals on the image data;

[0023] b22. Extract the gray values ​​of the reference line segment from bottom to top, use the distance from bottom to top as the category of the horizontal axis, and use the gray value at the corresponding distance as the category of the vertical axis, and then obtain the gray value change curve of the reference line segment.

[0024] b23. Extract the grayscale values ​​of the required feature parts from the historical data of the database, compare the grayscale value change curve with the grayscale values ​​of the required feature parts, record the initial and final points when the grayscale values ​​are the same, connect all the initial points in sequence according to the set order of the reference line segments to obtain the bottom boundary, and connect all the final points in sequence to obtain the top boundary. Combine the intersection of the bottom boundary and the top boundary with the vertical boundary of the image data to obtain the feature parts.

[0025] Preferably, the operation of comparing the material characteristic parameters with the material parameter thresholds in B3 to determine the stacking condition is as follows:

[0026] b31. The bottom boundary is horizontal. Extract the actual parameter threshold of the material and convert the actual parameter threshold into an image parameter threshold that matches the image data.

[0027] b32. A threshold line segment is set on the image data at a distance of the image parameter threshold from the bottom edge to the top edge. The material accumulation position is determined according to the intersection of the threshold line segment and the top edge.

[0028] b33. If the threshold line segment does not intersect with the top boundary, no material accumulation occurs. Conversely, if the threshold line segment intersects with the top boundary, material accumulation occurs in the area between the first and second intersection points, and material accumulation occurs in the area between every two subsequent intersection points.

[0029] Preferably, the operation in A2 to analyze and identify the image data to determine the features of the foreign object is as follows:

[0030] C1. After resolving the material accumulation issue, extract and analyze the image data collected by the color sorter.

[0031] C2. Extract material characteristic parameters from historical data in the database, and compare the combination of size and color parameters in the material characteristic parameters with the collected image data;

[0032] C3. Use edge algorithms to process the grayscale image data to determine the feature boundaries. Features whose size parameters after segmentation are greater than or less than the size parameter range in the historical data are considered foreign object features. Features whose color after segmentation is inconsistent with the color parameters in the historical data are also considered foreign object features.

[0033] Preferably, the operation in A2 to determine whether the foreign object feature is located in the screened material is as follows:

[0034] D1. Based on the foreign object characteristics identified in the image data, compare the image data collected after the movement at adjacent time stamps;

[0035] D2. Keep adjacent image data consistent, match the corner points of adjacent image data, and adjust the transparency of the upper image data so that the features of the covered lower image can also be displayed. Determine the location of foreign object features and perform contour comparison operation to determine that the foreign object features are located on the screening material or glass baffle.

[0036] Preferably, the contour comparison operation for determining the location of foreign object features in D2 is as follows:

[0037] d21. Identify foreign object features in the upper layer image data and determine whether the location of the foreign object feature corresponds to the location in the lower layer image data.

[0038] d22. When it is determined that both the upper and lower image data have foreign object features at the same location, the contour boundaries of the foreign object features are compared. If the contour boundaries of the two foreign object features are completely consistent, it is determined that the current foreign object feature is attached to the glass baffle. If the contour boundaries of the two foreign object features are different, the current foreign object feature is located in the screening material.

[0039] Preferably, the operation of generating control instructions according to the processing strategy in A3 is as follows:

[0040] a31. After determining the location of the material accumulation based on the accumulation situation, generate an instruction to smooth the material so that the accumulation thickness meets the threshold line segment requirements.

[0041] a32. When it is determined that the foreign object feature is adhered to the glass baffle, an instruction is generated to clean the foreign object feature adhered to the glass baffle.

[0042] a33. Determine that the foreign object characteristics are located in the screening material, and generate control instructions for screening the foreign object characteristics and the required material.

[0043] This invention provides a visual inspection method based on foreign object sieving in a color sorter. Compared with existing technologies, it has the following advantages:

[0044] 1. This visual inspection method for foreign object screening based on color sorters preprocesses the acquired image data, uses the image data to establish a three-dimensional model of material conveying to determine the material accumulation, analyzes and identifies foreign object characteristics in the image data to determine whether the foreign object characteristics are located in the screening material, and then generates a processing strategy based on the accumulation and foreign object characteristics for transmission. This method can more comprehensively acquire image data of the material to be screened, and ensure the efficiency and accuracy of visual inspection, thus better realizing the visual inspection operation of foreign object screening.

[0045] 2. This visual inspection method based on foreign object screening in a color sorter establishes a three-dimensional coordinate system, extracts image data collected at the same time stamp, analyzes and determines the feature parts, and then introduces the analyzed image data into the three-dimensional coordinate system according to the corresponding orientation to construct a three-dimensional model of material conveying. The material feature parameters in the three-dimensional model are compared with the material parameter thresholds to determine the accumulation situation, thereby effectively realizing the operation of handling accumulated materials and avoiding the problem of material information not being identified or being unclear during subsequent visual recognition, thus ensuring the integrity of visual inspection.

[0046] 3. This visual inspection method based on foreign object screening in color sorters determines the foreign object features in the upper image data and judges whether the corresponding position of the foreign object feature in the lower image data is a foreign object feature. The outline boundary of the foreign object feature is compared to determine whether the foreign object feature is adhered to the glass baffle or located in the screened material. This avoids the influence of external factors on visual inspection and improves the accuracy of visual inspection. Attached Figure Description

[0047] Figure 1 This is a flowchart illustrating the operation of the visual inspection method of the present invention. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] Please see Figure 1 This invention provides two technical solutions:

[0050] Example 1: A visual inspection method based on foreign object sieving in a color sorter, specifically including the following steps:

[0051] A1. The material to be screened is conveyed and image information of the conveyed material is collected. After the material to be screened enters the target area, the material to be screened is identified by a color sorter to obtain multiple image data containing the material to be screened. The obtained image data is transmitted and stored in the database.

[0052] A2. Preprocess the acquired image data, use the image data to build a three-dimensional model of material conveying to determine the material accumulation situation, analyze and identify the foreign object characteristics of the image data, and determine whether the foreign object characteristics are located in the screening material. Then, generate a processing strategy based on the accumulation situation and foreign object characteristics for transmission.

[0053] A3. Generate control commands based on the processing strategy and transmit them to the equipment units used for processing to complete the material processing operation.

[0054] By preprocessing the acquired image data, a three-dimensional model of material conveying is established using the image data to determine the material accumulation. The image data is then analyzed and identified to determine foreign object characteristics and whether these characteristics are located in the screening material. Based on the accumulation and foreign object characteristics, a processing strategy is generated for transmission. This approach enables more comprehensive acquisition of image data of the material to be screened, ensuring the efficiency and accuracy of visual inspection and better realizing the visual inspection operation for foreign object screening.

[0055] The equipment unit can be used to realize material conveying operations, and to smooth the accumulated materials so that the materials can be placed in a single layer to avoid the problem of covering. It can also clean foreign objects adhering to the glass baffle. At the same time, the instructions are transmitted to the color sorter to realize the screening operation of abnormal materials and required materials. The equipment unit is set up using existing mature technologies and relies on the control instructions generated by the system for transmission and control.

[0056] In this embodiment of the invention, the operation of image acquisition for conveying materials in A1 is as follows:

[0057] a11. Before entering the color sorter for identification, high-definition cameras are installed directly above and to the side of the conveyor table in the conveying stage to collect images;

[0058] a13. After entering the color sorter, a glass baffle is set between the color sorter and the material to be screened for separation, and the color sorter identifies the material to be screened below the glass baffle.

[0059] In this embodiment of the invention, the preprocessing operation for the acquired image data in A2 is as follows:

[0060] a21. Set up multi-level nodes and perform image content classification by matching the content of multi-level nodes with the parameter content of image data;

[0061] a22. Multi-level nodes are divided into first-level nodes, second-level nodes, and third-level nodes. The content of the first-level node is the image acquisition device information, the content of the second-level node is the orientation parameter information under the first-level node, and the content of the third-level node is the acquisition timestamp information under the second-level node.

[0062] a23. Match the parameter information associated with the collected image data with the content of the multi-level nodes. When the match is consistent, fill the image data into the corresponding node.

[0063] In this embodiment of the invention, the operation in A2 of establishing a three-dimensional model of material conveying using image data to determine the material accumulation is as follows:

[0064] B1. Establish a three-dimensional coordinate system. Set the X-axis and Y-axis at the plane where the conveyor is located. Take the point where a corner point of the image collected from the top of the image intersects with the boundary of the conveyor as the starting point. Establish the X-axis at the point where the starting point is parallel to the boundary of the conveyor. Establish the Y-axis perpendicular to the X-axis, intersecting at the starting point and extending to the opposite boundary of the conveyor. Establish the Z-axis vertically upward from the starting point.

[0065] B2. Extract image data collected at the same time stamp, analyze and determine the feature parts, and then import the analyzed image data into the three-dimensional coordinate system according to the corresponding orientation to construct a three-dimensional model of material conveying.

[0066] B3. Extract the parameter thresholds of the current material from the historical data in the database, and compare the material characteristic parameters in the 3D model with the parameter thresholds of the material to determine the stacking situation.

[0067] In this embodiment of the invention, the operation of analyzing image data and determining feature parts in B2 is as follows:

[0068] b21. Extract image data at the same timestamp, perform grayscale processing on the image data, and set multiple reference line segments parallel to the vertical boundary at equal intervals on the image data;

[0069] b22. Extract the gray values ​​of the reference line segment from bottom to top, use the distance from bottom to top as the category of the horizontal axis, and use the gray value at the corresponding distance as the category of the vertical axis, and then obtain the gray value change curve of the reference line segment.

[0070] b23. Extract the grayscale values ​​of the required feature parts from the historical data of the database, compare the grayscale value change curve with the grayscale values ​​of the required feature parts, record the initial and final points when the grayscale values ​​are the same, connect all the initial points in sequence according to the set order of the reference line segments to obtain the bottom boundary, and connect all the final points in sequence to obtain the top boundary. Combine the intersection of the bottom boundary and the top boundary with the vertical boundary of the image data to obtain the feature parts.

[0071] By establishing a three-dimensional coordinate system, image data collected at the same time stamp is extracted and analyzed to determine the feature parts. Then, the analyzed image data is introduced into the three-dimensional coordinate system according to the corresponding orientation to construct a three-dimensional model of material conveying. The material feature parameters in the three-dimensional model are compared with the material parameter thresholds to determine the accumulation situation. This effectively realizes the operation of handling accumulated materials and avoids the problem of material information not being recognized or being unclear during subsequent visual recognition, thus ensuring the integrity of visual inspection.

[0072] In this embodiment of the invention, the operation of comparing the material characteristic parameters in B3 with the material parameter thresholds to determine the stacking condition is as follows:

[0073] b31. The bottom boundary is horizontal. Extract the actual parameter threshold of the material and convert the actual parameter threshold into an image parameter threshold that matches the image data.

[0074] b32. A threshold line segment is set on the image data at a distance of the image parameter threshold from the bottom edge to the top edge. The material accumulation position is determined according to the intersection of the threshold line segment and the top edge.

[0075] b33. If the threshold line segment does not intersect with the top boundary, no material accumulation occurs. Conversely, if the threshold line segment intersects with the top boundary, material accumulation occurs in the area between the first and second intersection points, and material accumulation occurs in the area between every two subsequent intersection points.

[0076] In this embodiment of the invention, the operation of analyzing and identifying foreign object features in A2 is as follows:

[0077] C1. After resolving the material accumulation issue, extract and analyze the image data collected by the color sorter.

[0078] C2. Extract material characteristic parameters from historical data in the database, and compare the combination of size and color parameters in the material characteristic parameters with the collected image data;

[0079] C3. Use edge algorithms to process the grayscale image data to determine the feature boundaries. Features whose size parameters after segmentation are greater than or less than the size parameter range in the historical data are considered foreign object features. Features whose color after segmentation is inconsistent with the color parameters in the historical data are also considered foreign object features.

[0080] In this embodiment of the invention, the operation in A2 to determine whether the foreign object feature is located in the screened material is as follows:

[0081] D1. Based on the foreign object characteristics identified in the image data, compare the image data collected after the movement at adjacent time stamps;

[0082] D2. Keep adjacent image data consistent, match the corner points of adjacent image data, and adjust the transparency of the upper image data so that the features of the covered lower image can also be displayed. Determine the location of foreign object features and perform contour comparison operation to determine that the foreign object features are located on the screening material or glass baffle.

[0083] In this embodiment of the invention, the contour comparison operation for determining the location of the foreign object feature in D2 is as follows:

[0084] d21. Identify foreign object features in the upper layer image data and determine whether the location of the foreign object feature corresponds to the location in the lower layer image data.

[0085] d22. When it is determined that both the upper and lower image data have foreign object features at the same location, the contour boundaries of the foreign object features are compared. If the contour boundaries of the two foreign object features are completely consistent, it is determined that the current foreign object feature is attached to the glass baffle. If the contour boundaries of the two foreign object features are different, the current foreign object feature is located in the screening material.

[0086] By identifying foreign object features in the upper-layer image data and determining whether the corresponding location of the foreign object feature in the lower-layer image data is also a foreign object feature, the contour boundary of the foreign object feature is compared to determine whether the foreign object feature is adhering to the glass baffle or located in the screening material. This avoids the influence of external factors on visual inspection and improves the accuracy of visual inspection.

[0087] In this embodiment of the invention, the operation of generating control instructions according to the processing strategy in A3 is as follows:

[0088] a31. After determining the location of the material accumulation based on the accumulation situation, generate an instruction to smooth the material so that the accumulation thickness meets the threshold line segment requirements.

[0089] a32. When it is determined that the foreign object feature is adhered to the glass baffle, an instruction is generated to clean the foreign object feature adhered to the glass baffle.

[0090] a33. Determine that the foreign object characteristics are located in the screening material, and generate control instructions for screening the foreign object characteristics and the required material.

[0091] Example 2 differs from Example 1 in that it applies both the existing visual inspection method and the visual inspection method of this invention, which generates a processing strategy based on the packing situation and foreign object characteristics, to the same model of color sorter. The proportion of foreign object characteristics in the final sieving material and the proportion of the desired material within the foreign object characteristics are recorded. The specific results are shown in Table 1.

[0092] Table 1 Record Results Table

[0093]

[0094] In summary, the present invention provides a visual inspection method for generating processing strategies based on the stacking conditions and foreign object characteristics. When applied to a color sorter, the results show that both the proportion of foreign object characteristics in the desired material and the proportion of desired material in foreign object characteristics are lower than those of existing visual inspection methods. Therefore, the visual inspection method of the present invention is more effective and can be applied in practical operations.

[0095] Furthermore, any content not described in detail in this specification is existing technology known to those skilled in the art.

[0096] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0097] Although embodiments of the invention 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 to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A visual inspection method based on foreign object sieving in a color sorter, characterized in that: Specifically, the following steps are included: A1. The material to be screened is conveyed and image information of the conveyed material is collected. After the material to be screened enters the target area, the material to be screened is identified by a color sorter to obtain multiple image data containing the material to be screened. The obtained image data is transmitted and stored in the database. A2. Preprocess the acquired image data, use the image data to build a three-dimensional model of material conveying to determine the material accumulation situation, analyze and identify the foreign object characteristics of the image data, and determine whether the foreign object characteristics are located in the screening material. Then, generate a processing strategy based on the accumulation situation and foreign object characteristics for transmission. A3. Generate control commands based on the processing strategy and transmit them to the equipment units used for processing to complete the material processing operation; The operation in A2 that uses image data to build a three-dimensional model of material transport to determine the material accumulation is as follows: B1. Establish a three-dimensional coordinate system. Set the X-axis and Y-axis at the plane where the conveyor is located. Take the point where a corner point of the image collected from the top of the image intersects with the boundary of the conveyor as the starting point. Establish the X-axis at the point where the starting point is parallel to the boundary of the conveyor. Establish the Y-axis perpendicular to the X-axis, intersecting at the starting point and extending to the opposite boundary of the conveyor. Establish the Z-axis vertically upward from the starting point. B2. Extract image data collected at the same time stamp, analyze and determine the feature parts, and then import the analyzed image data into the three-dimensional coordinate system according to the corresponding orientation to construct a three-dimensional model of material conveying. B3. Extract the parameter thresholds of the current material from the historical data of the database, and compare the material characteristic parameters in the 3D model with the parameter thresholds of the material to determine the stacking situation; The operation in B2 to analyze image data and determine feature parts is as follows: b21. Extract image data at the same timestamp, perform grayscale processing on the image data, and set multiple reference line segments parallel to the vertical boundary at equal intervals on the image data; b22. Extract the gray values ​​of the reference line segment from bottom to top, use the distance from bottom to top as the category of the horizontal axis, and use the gray value at the corresponding distance as the category of the vertical axis, and then obtain the gray value change curve of the reference line segment. b23. Extract the gray values ​​of the required feature parts from the historical data of the database, compare the gray value change curve with the gray values ​​of the required feature parts, record the initial point and the end point when the gray values ​​are the same, connect all the initial points in sequence according to the set order of the reference line segments to obtain the bottom boundary, and connect all the end points in sequence to obtain the top boundary. Combine the intersection of the bottom boundary and the top boundary with the vertical boundary of the image data to obtain the feature part. The operation in B3 that compares the material characteristic parameters with the material parameter thresholds to determine the stacking condition is as follows: b31. The bottom boundary is horizontal. Extract the actual parameter threshold of the material and convert the actual parameter threshold into an image parameter threshold that matches the image data. b32. A threshold line segment is set on the image data at a distance of the image parameter threshold from the bottom boundary to the top boundary. The material accumulation position is determined according to the intersection of the threshold line segment and the top boundary. b33. If the threshold line segment does not intersect with the top boundary, no material accumulation occurs. Conversely, if the threshold line segment intersects with the top boundary, material accumulation occurs in the area between the first and second intersection points, and material accumulation occurs in the area between every two subsequent intersection points.

2. The visual inspection method based on foreign object sieving in a color sorter according to claim 1, characterized in that: The operation of image acquisition for conveyed materials in A1 is as follows: a11. Before entering the color sorter for identification, high-definition cameras are installed directly above and to the side of the conveyor table in the conveying stage to collect images; a13. After entering the color sorter, a glass baffle is set between the color sorter and the material to be screened for separation, and the color sorter identifies the material to be screened below the glass baffle.

3. The visual inspection method based on foreign object sieving in a color sorter according to claim 1, characterized in that: The preprocessing operation for the acquired image data in A2 is as follows: a21. Set up multi-level nodes and perform image content classification by matching the content of multi-level nodes with the parameter content of image data; a22. Multi-level nodes are divided into first-level nodes, second-level nodes, and third-level nodes. The content of the first-level node is the image acquisition device information, the content of the second-level node is the orientation parameter information under the first-level node, and the content of the third-level node is the acquisition timestamp information under the second-level node. a23. Match the parameter information associated with the collected image data with the content of the multi-level nodes. When the match is consistent, fill the image data into the corresponding node.

4. The visual inspection method based on foreign object sieving in a color sorter according to claim 1, characterized in that: The operation in A2 to analyze and identify image data to determine foreign object characteristics is as follows: C1. After resolving the material accumulation issue, extract and analyze the image data collected by the color sorter. C2. Extract material characteristic parameters from historical data in the database, and compare the combination of size and color parameters in the material characteristic parameters with the collected image data; C3. Use edge algorithms to process the grayscale image data to determine the feature boundaries. Features whose size parameters after segmentation are greater than or less than the size parameter range in the historical data are considered foreign object features. Features whose color after segmentation is inconsistent with the color parameters in the historical data are also considered foreign object features.

5. The visual inspection method based on foreign object sieving in a color sorter according to claim 1, characterized in that: The operation in A2 to determine whether foreign object characteristics are located in the screened material is as follows: D1. Based on the foreign object characteristics identified in the image data, compare the image data collected after the movement at adjacent time stamps; D2. Keep adjacent image data consistent, match the corner points of adjacent image data, and adjust the transparency of the upper image data so that the features of the covered lower image can also be displayed. Determine the location of foreign object features and perform contour comparison operation to determine that the foreign object features are located on the screening material or glass baffle.

6. The visual inspection method based on foreign object sieving in a color sorter according to claim 5, characterized in that: The contour comparison operation for determining the location of foreign object features in D2 is as follows: d21. Identify foreign object features in the upper layer image data and determine whether the location of the foreign object feature corresponds to the location in the lower layer image data. d22. When it is determined that both the upper and lower image data have foreign object features at the same location, the contour boundaries of the foreign object features are compared. If the contour boundaries of the two foreign object features are completely consistent, it is determined that the current foreign object feature is attached to the glass baffle. If the contour boundaries of the two foreign object features are different, the current foreign object feature is located in the screening material.

7. A visual inspection method based on foreign object sieving in a color sorter according to claim 5, characterized in that: The operation in A3 that generates control instructions according to the processing strategy is as follows: a31. After determining the location of the material accumulation based on the accumulation situation, generate an instruction to smooth the material so that the accumulation thickness meets the threshold line segment requirements. a32. When it is determined that the foreign object feature is adhered to the glass baffle, an instruction is generated to clean the foreign object feature adhered to the glass baffle. a33. Determine that the foreign object characteristics are located in the screening material, and generate control instructions for screening the foreign object characteristics and the required material.

Images