A burr detection method and system for copper rod electrolytic polishing

Abstract

The present invention relates to the technical field of copper rod burr detection, and in particular to a method and system for detecting burrs in copper rod electrolytic polishing. The method comprises: first, continuously collecting copper rod surface images by determining the movement direction of the copper rod, and performing dynamic feature analysis on the copper rod surface images based on the collected physical properties of the electrolyte to obtain an electrolyte interference area, and then comparing and identifying the electrolyte interference area with a plurality of copper rod surface images to obtain a burr candidate area, and then screening the burr candidate area based on an established copper rod polishing database to obtain a burr detection result. The present invention effectively solves the problem that burr detection is easily interfered with by human factors, thereby failing to provide efficient and stable detection results, improves the accuracy and reliability of burr detection, and thus significantly improves the detection efficiency and product quality of copper rod electrolytic polishing.

Description

Technical Field

[0001] The present invention relates to the technical field of copper rod burr detection, and in particular to a copper rod electrolytic polishing burr detection method and system. Background Art

[0002] With the rapid development of modern manufacturing and the increasing demand for precision machining, copper rod electropolishing technology has gained widespread application in the electrical and electronic fields. Electropolishing not only improves the surface finish of copper rods but also effectively enhances their corrosion resistance and electrical conductivity, making it crucial in the production of many high-precision products. However, during the electropolishing process, burrs are prone to forming on the copper rod surface, impacting subsequent processing and product quality. The presence of burrs can not only affect the copper rod's appearance but also lead to unstable electrical conductivity or poor connections, thereby compromising the reliability and performance of the entire electronic product.

[0003] However, existing burr detection methods mostly rely on manual visual inspection or simple mechanical inspection equipment. These traditional methods are not only inefficient but also easily interfered with by human factors and cannot provide efficient and stable detection results. In addition, the dynamic interference caused by electrolyte flow is highly similar to the optical characteristics of real burrs. Existing technologies have not yet been able to provide a systematic and accurate solution in terms of electrolyte interference, dynamic surface changes and burr feature identification, which limits the application potential of burr detection technology in actual production.

[0004] The information disclosed in this background technology section is only intended to deepen the understanding of the overall background technology of the present disclosure and should not be regarded as an admission or any form of suggestion that the information constitutes the prior art known to those skilled in the art. Summary of the Invention

[0005] The present invention provides a method and system for detecting burrs in electrolytic polishing of a copper rod, which can effectively solve the problems in the background technology.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A method for detecting burrs during electrolytic polishing of a copper rod, the method comprising:

[0008] Determine the moving direction of the copper rod, and continuously collect a plurality of copper rod surface images according to the moving direction of the copper rod, wherein the plurality of copper rod surface images are multi-frame time-series images;

[0009] Collecting physical properties of the electrolyte, performing dynamic feature analysis on the copper rod surface images of adjacent frames based on the physical properties of the electrolyte, and identifying and screening electrolyte interference areas;

[0010] Comparing and identifying the electrolyte interference area with a plurality of copper rod surface images respectively to obtain a burr candidate area;

[0011] A copper rod polishing database is established, burr static features are extracted according to the copper rod polishing database, and the burr candidate areas are screened according to the burr static features to obtain a burr detection result.

[0012] Furthermore, dynamic feature analysis is performed on the copper rod surface images of the adjacent frames according to the physical properties of the electrolyte to identify and screen the electrolyte interference area, including:

[0013] Preprocessing the copper rod surface images of adjacent frames, performing pixel point displacement calculation on the preprocessed copper rod surface images to obtain original pixel displacement;

[0014] Collecting copper rod displacement data according to the moving direction of the copper rod, and performing motion compensation on the original pixel displacement according to the copper rod displacement data to obtain electrolyte displacement;

[0015] Verifying the electrolyte displacement according to the physical properties of the electrolyte to generate an interference marker matrix;

[0016] The topological relationship of the interference marker matrix is reconstructed to obtain the electrolyte interference area.

[0017] Furthermore, topological relationship reconstruction is performed on the interference marker matrix to obtain the electrolyte interference area, including:

[0018] Merging adjacent interference mark points in the interference mark matrix into continuous regions to obtain an initial interference region set;

[0019] Filling each initial interference region hole in the initial interference region set and smoothing the region edge to generate an optimized interference region set, wherein the initial interference region hole is a discontinuous blank region in the initial interference region;

[0020] Setting an interference area threshold according to the physical properties of the electrolyte, and eliminating the optimized interference region set according to the interference area threshold;

[0021] The optimized interference region set after elimination is subjected to liquid morphology verification, and the region that meets the flow characteristics of the electrolyte is marked as the electrolyte interference region.

[0022] Furthermore, each initial interference region hole in the initial interference region set is filled, and the region edge is smoothed to generate an optimized interference region set, including:

[0023] identifying closed holes based on the initial interference area, wherein the closed holes represent areas completely surrounded by the initial interference area;

[0024] collecting electrolysis process parameters in real time, determining the hole filling size according to the physical characteristics of the electrolyte, and determining the expanded pixel value based on the electrolysis process parameters;

[0025] The hole boundary is expanded according to the hole filling size and the expanded pixel value, and the closed hole after expansion is smoothed to obtain an optimized interference area.

[0026] Furthermore, the electrolyte interference area is compared and identified with a plurality of copper rod surface images to obtain a burr candidate area, including:

[0027] Marking and removing the electrolyte interference area according to the copper rod surface image to obtain an effective detection area;

[0028] Acquire spectral imaging data of the copper rod surface image, and calculate and acquire an optical band response difference of each pixel in the effective detection area based on the spectral imaging data, wherein the optical band response difference represents a difference in reflectivity of a pixel point for two different bands;

[0029] determining a difference direction according to the optical band response difference, and marking potential burr pixels according to the difference direction;

[0030] Boundaries of the potential burr pixels are closed, physical features are verified based on the burr pixels after boundary closure, and burr candidate areas are marked based on the physical feature verification results.

[0031] Furthermore, the static features of the burr are extracted, including:

[0032] Acquire a historical burr detection image according to the copper rod polishing database, and perform grayscale conversion on the historical burr detection image to acquire a grayscale image;

[0033] Preprocessing the grayscale image, and calculating the gradient magnitude and gradient direction based on the preprocessed grayscale image;

[0034] Obtaining an angle threshold and a gradient amplitude threshold according to the copper rod polishing database, screening the angle between the gradient direction and the copper rod moving direction according to the angle threshold, obtaining a gradient response area, and dividing the gradient response area into a burr feature area and a potential burr feature area according to the gradient amplitude threshold;

[0035] Performing correlation verification of the geometric shape and spectral reflectance characteristics of the potential burr feature area, and connecting with the burr feature area if a preset verification condition is met;

[0036] Static features are extracted from the connected burr feature areas to obtain a burr static feature set.

[0037] Furthermore, the gradient magnitude and gradient direction are calculated, including:

[0038] Performing convolution operations on the grayscale image in the horizontal direction and the vertical direction respectively to obtain convolution operation results;

[0039] Collecting the actual surface curvature of the copper rod and the surface curvature of the image, obtaining an axial deformation compensation factor, and calculating the gradient amplitude of the pixel point according to the convolution operation result and the axial deformation compensation factor;

[0040] According to the vector relationship between the convolution operation results in the horizontal and vertical directions, the gradient direction of each pixel point is calculated and obtained.

[0041] Furthermore, marking the burr candidate area according to the physical feature verification result and optimizing the burr candidate area includes:

[0042] Perform binary segmentation on the burr candidate area based on a deep learning algorithm to obtain a binary segmentation image;

[0043] Extracting burr static features according to the copper rod polishing database, and constructing a sensitive convolution kernel according to the burr static features;

[0044] The binary segmentation image is edge-filled and denoised according to the sensitive convolution kernel, and linear fitting is performed on the binary segmentation image after the edge-filled and denoised processing to optimize the burr candidate area.

[0045] A burr detection system for copper rod electrolytic polishing, the system comprising:

[0046] A copper rod image acquisition module determines the moving direction of the copper rod and continuously acquires a plurality of copper rod surface images according to the moving direction of the copper rod, wherein the plurality of copper rod surface images are multi-frame time-series images;

[0047] An electrolyte interference screening module collects physical properties of the electrolyte, performs dynamic feature analysis on the copper rod surface images of adjacent frames based on the physical properties of the electrolyte, and identifies and screens electrolyte interference areas;

[0048] a burr candidate identification module, which compares and identifies the electrolyte interference area with a plurality of copper rod surface images to obtain a burr candidate area;

[0049] The burr feature screening module establishes a copper rod polishing database, extracts burr static features according to the copper rod polishing database, and screens the burr candidate areas according to the burr static features to obtain burr detection results.

[0050] Furthermore, the electrolyte interference screening module includes:

[0051] An image preprocessing calculation unit preprocesses the copper rod surface images of adjacent frames, performs pixel point displacement calculation on the preprocessed copper rod surface images, and obtains original pixel displacement;

[0052] a motion compensation processing unit, which collects copper rod displacement data according to the moving direction of the copper rod, and performs motion compensation on the original pixel displacement according to the copper rod displacement data to obtain electrolyte displacement;

[0053] an interference verification marking unit, which verifies the displacement of the electrolyte according to the physical properties of the electrolyte and generates an interference marking matrix;

[0054] The interference region reconstruction unit reconstructs the topological relationship of the interference mark matrix to obtain the electrolyte interference region.

[0055] The technical solution of the present invention can achieve the following technical effects:

[0056] It effectively solves the problem of difficulty in identification due to the high similarity in optical characteristics between the dynamic interference caused by the flow of electrolyte and the real burrs. Through comprehensive analysis of the copper rod surface image and the physical properties of the electrolyte, it can accurately identify and screen the electrolyte interference area, effectively avoid the influence of electrolyte interference on the burr detection results, improve the accuracy of burr detection, enhance the surface quality of the copper rod, optimize the production process, and reduce the defective rate.

[0057] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0059] Figure 1 The figure is a flow chart of a burr detection method for electrolytic polishing of a copper rod;

[0060] Figure 2 Schematic diagram of the process for identifying electrolyte interference areas;

[0061] Figure 3Schematic diagram of the process of topological reconstruction of interference marking matrix;

[0062] Figure 4 Schematic diagram of the topological reconstruction of the interference marker matrix. DETAILED DESCRIPTION

[0063] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0064] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains. The terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0065] Embodiment 1;

[0066] like Figure 1 As shown, the present application provides a method for detecting burrs in electrolytic polishing of a copper rod, the method comprising:

[0067] S10: determining the moving direction of the copper rod, and continuously acquiring a plurality of copper rod surface images according to the moving direction of the copper rod, wherein the plurality of copper rod surface images are multi-frame time-series images;

[0068] S20: Collecting the physical properties of the electrolyte, performing dynamic feature analysis on the copper rod surface images of adjacent frames based on the physical properties of the electrolyte, and identifying and screening the electrolyte interference area;

[0069] S30: comparing and identifying the electrolyte interference area with several copper rod surface images to obtain burr candidate areas;

[0070] S40: establishing a copper rod polishing database, extracting burr static features based on the copper rod polishing database, and screening burr candidate areas based on the burr static features to obtain burr detection results.

[0071] Specifically, first, determine the moving direction of the copper rod during the electrolytic polishing process. In some embodiments, an encoder can be used to monitor the rotation or movement angle of the copper rod in real time. The encoder can provide accurate synchronization information for subsequent image acquisition by accurately measuring the rotation or linear displacement of the copper rod. A photoelectric sensor installed on the polishing machine can also be used to detect the displacement of the copper rod to determine the movement direction of the copper rod. According to the determined moving direction of the copper rod, a number of copper rod surface images are continuously collected. These images are multi-frame time-series images, and each frame of the image corresponds to a time point. During the image acquisition process, a high-speed industrial camera can be used to capture the image. The acquisition frequency should be adjusted according to the moving speed of the copper rod. In order to ensure the integrity and details of the image data, a high-resolution camera can be used. While acquiring the image, the physical properties of the electrolyte (such as flow rate, temperature, concentration and viscosity, etc.) need to be collected in real time. ; The physical properties of the electrolyte have an important influence on the copper rod polishing process, especially the flow rate and temperature, which will directly affect the flow of the electrolyte and the polishing effect of the copper rod surface. For this reason, special electrolyte physical property sensors are used, such as flow meters, temperature sensors and concentration sensors, to ensure real-time data acquisition; after collecting the physical properties of the electrolyte, the surface images of the copper rod in adjacent frames are then subjected to dynamic feature analysis. Dynamic feature analysis refers to extracting features related to time changes by analyzing the changes between consecutive frames, so as to accurately identify dynamic elements in the image; in the present invention, the purpose of dynamic feature analysis is to capture and analyze the interference areas caused by electrolyte flow, bubbles or foam from the surface images of the copper rod in adjacent frames; in this way, the difference between these interference factors and the real features of the copper rod surface can be effectively distinguished, thereby identifying and screening the electrolyte interference areas. Then, the screened electrolyte interference area is compared with the copper rod surface image to obtain the burr candidate area. In order to improve the accuracy of burr detection, this method uses the historical burr data in the copper rod polishing database to extract static features. The copper rod polishing database contains burr feature data under different process conditions and different electrolyte parameters, including burr size, shape, edge smoothness, etc. Then, the burr candidate area is screened according to the burr static features, and finally the burr detection result is obtained.

[0072] The technical solution of the present invention effectively solves the problem of difficulty in identification caused by the dynamic interference generated by the electrolyte flow and the real burrs being highly similar in optical characteristics. Through comprehensive analysis of the copper rod surface image and the physical properties of the electrolyte, the electrolyte interference area can be accurately identified and screened, effectively avoiding the influence of electrolyte interference on the burr detection results, improving the accuracy of burr detection, enhancing the surface quality of the copper rod, optimizing the production process, and reducing the defective product rate.

[0073] Further, if Figure 2As shown, the copper rod surface images of adjacent frames are dynamically analyzed based on the physical properties of the electrolyte to identify and screen the electrolyte interference area, including:

[0074] S21: Preprocessing the copper rod surface images of adjacent frames, performing pixel point displacement calculation on the preprocessed copper rod surface images to obtain original pixel displacement;

[0075] S22: collecting copper rod displacement data according to the moving direction of the copper rod, and performing motion compensation on the original pixel displacement according to the copper rod displacement data to obtain the electrolyte displacement;

[0076] S23: Verify the electrolyte displacement according to the physical properties of the electrolyte and generate an interference marker matrix;

[0077] S24: Reconstruct the topological relationship of the interference marker matrix to obtain the electrolyte interference area.

[0078] As a preferred embodiment of the above, the collected copper rod surface image is first preprocessed, and the preprocessing includes steps such as denoising, image sharpening, and contrast enhancement, so as to improve the image quality and provide clearer image data for the subsequent pixel displacement calculation; since the copper rod surface will move or displace to a certain extent over time in the continuously collected adjacent frame images, this will cause errors in the actual measurement of the electrolyte displacement, so it is necessary to perform pixel point displacement calculation on the preprocessed copper rod surface image, and the method is as follows: first, the adjacent frames are aligned, which can be achieved by a feature-based image alignment algorithm (such as SIFT, SURF or ORB algorithm) to find the matching points between the images, and then the pixel displacement is calculated by the optical flow method. As follows: Assume that the brightness of the object is constant between frames that are very short in time. Then, by analyzing the brightness changes of each pixel in adjacent frames and comparing the difference in the brightness of each pixel in the two frames, the displacement of the pixel is inferred, thereby obtaining the original pixel displacement. The movement of the pixel can be represented by a velocity vector, which describes the position change of the pixel from the previous frame to the current frame. Moreover, since the copper rod moves during the electrolytic polishing process, the electrolyte flow may also affect the physical features in the image. Therefore, it is necessary to perform motion compensation on the original pixel displacement. During the compensation process, the system will adjust the original pixel displacement according to the displacement data of the copper rod (provided by the motion sensor) to eliminate the influence of the displacement of the copper rod itself on the image analysis results, and finally obtain the electrolyte displacement. Because the physical properties of the electrolyte, such as flow rate and temperature, affect its flow on the copper rod surface, the electrolyte displacement must be verified. In some embodiments, a fast matching method can be used to divide the copper rod surface image into multiple small blocks and compare them with the corresponding areas in the next time frame. By calculating the similarity between these small blocks, the system can infer the displacement of each small block. In particular, on the copper rod surface, the flow of electrolyte will cause pixels in certain areas to change. The system uses block matching to accurately track these changes, identify and mark interference areas caused by electrolyte flow or bubbles. Then, by comparing the displacement changes caused by electrolyte flow, the system can identify interference areas in the image caused by factors such as electrolyte flow, bubbles or foam. These interference areas usually appear as local discontinuities or irregularities in the image, especially in areas with fast or unstable flow, which may cause large displacements between pixels. After identifying these areas, the system generates an interference mark matrix, where each element of the matrix indicates whether the corresponding position in the image is an interference area.If a pixel position is affected by electrolyte flow or bubbles, its corresponding matrix element will be marked as interference (for example, marked as 1); if it is not affected, the element is marked as non-interference (for example, marked as 0). The generated interference mark matrix can be used to filter out areas in the image affected by factors such as electrolyte flow and bubbles. These areas are usually eliminated in subsequent steps to avoid unnecessary interference with burr detection. Finally, the topological relationship of the interference mark matrix is reconstructed to obtain the electrolyte interference area.

[0079] Further, if Figure 3 and Figure 4 As shown, the topological relationship of the interference marker matrix is reconstructed to obtain the electrolyte interference area, including:

[0080] S241: Merge adjacent interference marker points in the interference marker matrix into continuous regions to obtain an initial interference region set;

[0081] S242: Fill each initial interference region hole in the initial interference region set and smooth the region edge to generate an optimized interference region set, where the initial interference region hole is a discontinuous blank region in the initial interference region;

[0082] S243: setting an interference area threshold according to the physical properties of the electrolyte, and eliminating the optimized interference region set according to the interference area threshold;

[0083] S244: Perform liquid morphology verification on the set of optimized interference regions after elimination, and mark the regions that meet the flow characteristics of the electrolyte as electrolyte interference regions.

[0084] In this embodiment, the system will first traverse each pixel in the interference marker matrix, and for each interference marker (for example, marked as "1"), the system will check the pixels in its eight neighborhoods. If any pixel in these neighborhoods is also marked as "1", the system will consider that the two pixels are connected and they belong to the same interference area. At this time, the system will merge the two pixels, and the merged markers will be assigned the same area number or label to form a connected area, and the initial interference area set will be obtained. Next, each area in the initial interference area set is further processed to check whether there are holes in each initial interference area (that is, non-continuous blank areas in the area). If there are holes, the "hole filling" technology in image processing is usually used to fill these blank areas; and the edges of each area are smoothed to remove jagged edges caused by electrolyte interference or image noise, thereby generating an optimized interference area set. After that, the optimized interference region set will be further screened based on the physical properties of the electrolyte. The threshold of the interference area will be set and judged according to the physical properties of the electrolyte, such as the flow rate and temperature. If the area of a certain interference region is smaller than the set threshold, the region will be considered an invalid region and removed from the optimized interference region set. Through this screening step, the system can effectively remove smaller interference regions that may be caused by bubbles or small disturbances, reducing false detections. After removing the invalid regions, the optimized interference region set is further subjected to liquid morphology verification. Liquid morphology verification refers to the detection and verification of the morphological characteristics of the electrolyte interference region to determine whether these regions conform to the natural characteristics of the electrolyte flow. The purpose is to ensure that the identified interference region is caused by factors such as electrolyte flow, bubbles or foam, rather than other noise or incorrectly marked interference regions. The liquid morphology verification method can refer to the following method: morphological analysis (such as corrosion or expansion) is performed based on the electrolyte flow characteristics (such as fluidity, morphology, etc.). This analysis will take into account the movement trajectory of the electrolyte on the copper rod surface to ensure that the detected interference region conforms to the actual characteristics of the electrolyte flow. The region that conforms to the flow characteristics will be marked as the electrolyte interference region.

[0085] Furthermore, each initial interference region hole in the initial interference region set is filled and the region edges are smoothed to generate an optimized interference region set, including:

[0086] Identify closed holes based on the initial interference area, where closed holes represent areas completely surrounded by the initial interference area;

[0087] Real-time collection of electrolysis process parameters, determination of hole filling size based on the physical characteristics of the electrolyte, and determination of expanded pixel values based on the electrolysis process parameters;

[0088] The hole boundary is expanded according to the hole filling size and the expanded pixel value, and the closed hole after expansion is smoothed to obtain the optimized interference area.

[0089] Specifically, in the initial interference area set, image processing methods (such as corrosion and expansion in morphological operations) can be used to identify closed holes. Closed holes refer to blank areas that are completely surrounded by the initial interference area. These areas are usually caused by uneven electrolyte flow, bubbles or other external factors. After that, the electrolysis process parameters are collected in real time. These parameters include electrolyte flow rate, temperature, and current density, which have a direct impact on the flow characteristics of the electrolyte on the copper rod surface. Then, based on the physical characteristics of the electrolyte, especially factors such as flow rate and fluidity, the filling size of each hole is calculated. The filling size calculation method can be as follows: the area of the hole is calculated based on the pixel value of the area. The filling size is proportional to the area of the hole. The electrolyte fluidity also affects the filling size. The electrolyte with stronger fluidity can quickly fill larger holes. Therefore, based on the value of the electrolyte fluidity, the filling size is adjusted. For example, if the fluidity is strong (the viscosity is low), a larger filling size will be used to simulate the rapid filling of the electrolyte flow. The extended pixel value is calculated according to the electrolysis process parameters (including electrolyte fluidity, electrolyte temperature and electrolyte current density, etc.). The calculation formula is as follows:

[0090] ;

[0091] in, Indicates the extended pixel value, indicating the pixel size of the hole expansion. Indicates the electrolyte fluidity, represents the electrolyte temperature, Indicates standard temperature, is the current density of the electrolyte, represents the standard current density, Represents the adjustment coefficient, which is obtained based on historical experience. The hole boundary is then expanded based on the hole filling size and the expanded pixel value. A smoothing algorithm (such as Gaussian filtering) is applied to the expanded hole boundary to blend it with the surrounding area edges to obtain the optimized interference area.

[0092] Furthermore, the electrolyte interference area is compared with several copper rod surface images to identify candidate burr areas, including:

[0093] According to the copper rod surface image, the electrolyte interference area is marked and removed to obtain the effective detection area;

[0094] Acquire spectral imaging data of the copper rod surface image, and calculate and obtain the optical band response difference of each pixel in the effective detection area based on the spectral imaging data. The optical band response difference represents the difference in reflectivity of the pixel point for two different bands;

[0095] Determine the difference direction according to the difference in optical band response, and mark potential glitch pixels according to the difference direction;

[0096] The boundaries of potential burr pixels are closed, physical features are verified based on the burr pixels after boundary closure, and the burr candidate areas are marked based on the physical feature verification results.

[0097] As a preferred embodiment of the above embodiment, the electrolyte interference area is first removed from the copper rod surface image to obtain an effective detection area. Within the effective detection area, spectral imaging technology can be used to collect multi-band reflectance data for each pixel. Multi-band reflectance refers to the light reflectance ability of an object's surface at multiple different wavelengths (or bands). Different materials have different reflective properties for light in different bands. Therefore, by measuring reflectance in multiple bands, rich information about the composition, structure, and properties of the object's surface can be obtained. Then, by comparing the reflectance between different spectral bands, the band reflectance difference of each pixel is calculated to obtain an optical band response difference. This difference represents the difference in reflectance of the pixel in different bands. This difference can be used as an effective feature for identifying burrs. Because the optical reflectance characteristics between burrs and surrounding surfaces differ significantly, they are significantly highlighted in the difference in response between different bands. Based on the calculated optical band response difference, these difference directions are analyzed. The optical difference direction generally refers to the direction or band in which the reflectance difference of a specific pixel point in the spectral data of multiple bands is primarily reflected. In terms of emissivity changes, in short, the optical difference direction describes the directionality of the reflectivity difference between different spectral bands, indicating which band among multiple spectral bands contributes the most to the reflectivity difference of the pixel; by calculating the response difference of each band, the direction with the largest or most obvious difference is found in the entire image. This direction can be used as a basis for distinguishing areas such as burrs and stains, thereby obtaining potential burr pixels; then morphological operations such as corrosion and dilation can be used to eliminate breaks or gaps caused by image noise or interference, fill the gaps in the burr area, and thus form a complete burr candidate area.

[0098] Furthermore, the static features of the burr are extracted, including:

[0099] Acquire a historical burr detection image based on a copper rod polishing database, and perform grayscale conversion on the historical burr detection image to obtain a grayscale image;

[0100] Preprocessing the grayscale image, and calculating the gradient magnitude and gradient direction based on the preprocessed grayscale image;

[0101] An angle threshold and a gradient amplitude threshold are obtained based on the copper rod polishing database. The angle between the gradient direction and the copper rod moving direction is screened based on the angle threshold to obtain a gradient response area. The gradient response area is then divided into a burr feature area and a potential burr feature area based on the gradient amplitude threshold.

[0102] Perform correlation verification on the geometric shape and spectral reflectance characteristics of the potential burr feature area. If the preset verification conditions are met, it will be connected to the burr feature area;

[0103] Extract static features from the connected burr feature area to obtain a burr static feature set.

[0104] In this example, a historical burr detection image is first obtained from a copper rod polishing database. The RGB channel values of each pixel in the historical burr detection image are extracted and weighted averaged to obtain grayscale values for several pixel values. These grayscale values are then integrated to generate a grayscale image. To improve the accuracy of subsequent detection, the system preprocesses the grayscale image. Preprocessing steps typically include noise removal, image enhancement, and contrast adjustment to highlight the details of the burr area. After preprocessing, the image's gradient amplitude and gradient direction are calculated based on the image's pixel values. The gradient amplitude is the intensity of the change in pixel value in the image, reflecting the speed of surface change, while the gradient direction indicates the direction of the pixel value change. After calculating the gradient amplitude and gradient direction, an angle threshold is set. For example, an angle range of ±15° or ±30° is set. If the angle between the gradient direction of a pixel and the copper rod's movement direction falls within this range, the area is considered aligned with the copper rod's movement direction, indicating the presence of a burr. Only when the angle between the gradient direction and the copper rod's movement direction is less than the preset angle threshold is the pixel considered to have a strong response, indicating the presence of a possible burr. The gradient amplitude reflects the intensity of the brightness change within the area. On the copper rod's surface, the presence of burrs typically causes dramatic changes in surface brightness, which manifests as a large gradient amplitude. The system sets a minimum gradient amplitude threshold based on the brightness distribution of the copper rod surface image. For example, a gradient amplitude threshold is set to 5 or 10, depending on the image resolution and the desired response intensity. Combining the results of gradient direction screening and gradient amplitude screening, the system will mark pixel areas where the gradient direction is aligned with the copper rod movement direction and the gradient amplitude is greater than the threshold as potential burr feature areas. The potential burr area is then verified by correlating the geometric morphology with the spectral reflectance characteristics. The purpose of the geometric morphology verification is to determine whether the shape characteristics of the potential burr area meet the common geometric characteristics of burrs. Burrs typically have specific geometric morphologies, such as sharp edges, protruding structures, or irregular shapes. In some embodiments, geometric morphology verification is typically performed using morphological analysis. The purpose of the spectral reflectance characteristic verification is to analyze the reflectance characteristics of the potential burr area through spectral imaging data to confirm whether the area is a burr. The reflection differences of these areas in different spectral bands are calculated to confirm the nature of the burr. The geometric morphology verification results are combined with the spectral reflectance characteristic results to comprehensively judge the authenticity of the potential burr area. If the geometric shape of a potential burr area (such as sharp edges and protrusions) and its reflectance characteristics in multiple spectral bands meet the preset standards, then the area is considered to be an actual burr area; after the geometric shape and spectral reflectance characteristics of the potential burr area are verified, it is connected to the burr feature area. These connected burr feature areas will further extract static features. The static feature extraction process includes calculating various geometric features of the burr area, such as the area, shape, edge smoothness, etc., to obtain a set of burr static features.

[0105] Furthermore, the gradient magnitude and gradient direction are calculated, including:

[0106] Perform convolution operations on the grayscale image in the horizontal direction and the vertical direction respectively to obtain the convolution operation results;

[0107] The actual surface curvature of the copper rod and the image surface curvature are collected to obtain the axial deformation compensation factor. The gradient amplitude of the pixel point is calculated based on the convolution operation result and the axial deformation compensation factor.

[0108] According to the vector relationship between the convolution operation results in the horizontal and vertical directions, the gradient direction of each pixel is calculated and obtained.

[0109] Specifically, the Sobel operator can be used to convolve the image in the horizontal and vertical directions respectively. The convolution operations in these two directions can respectively extract the features of the horizontal and vertical edges in the image, thereby obtaining the local gradient information of each pixel in the horizontal and vertical directions. The convolution operation result will provide a horizontal and vertical gradient value for each pixel in the image; then the horizontal and vertical values of the convolution operation results are combined to calculate the original gradient amplitude of each pixel. The original gradient amplitude represents the intensity of the brightness change of a point in the image. The calculation formula is:

[0110] ;

[0111] in, and It is the gradient value in the horizontal and vertical directions; since the surface of the copper rod may be affected by physical deformation, such as surface curvature or irregularity, in order to eliminate the influence of these deformations on the gradient calculation, compensation is also performed by collecting the actual surface curvature data of the copper rod: using a surface curvature sensor or image analysis algorithm to measure the curvature of the copper rod surface, and calculate the deformation compensation factor based on the surface curvature data, and adjust the original gradient amplitude according to the compensation factor to eliminate the error caused by the surface curvature and ensure the accuracy of the gradient amplitude and gradient direction; then according to the gradient values in the horizontal and vertical directions,

[0112] Calculate the gradient direction of each pixel using the following formula:

[0113] ;

[0114] in, and are the gradient values in the horizontal and vertical directions.

[0115] Furthermore, the burr candidate areas are marked based on the physical feature verification results, and the burr candidate areas are optimized, including:

[0116] Perform binary segmentation on the burr candidate area based on the deep learning algorithm to obtain a binary segmentation image;

[0117] The static features of burrs are extracted based on the copper rod polishing database, and the sensitive convolution kernel is constructed based on the static features of burrs.

[0118] The binary segmentation image is edge-filled and denoised according to the sensitive convolution kernel, and linear fitting is performed on the binary segmentation image after edge-filling and denoising to optimize the burr candidate area.

[0119] As a preferred embodiment of the above, a deep learning algorithm is first used to perform binary segmentation on the marked burr candidate area. The purpose of binary segmentation is to separate the burr area in the image from the background. Specifically, a convolutional neural network can be trained. The network has been trained on a large number of copper rod burr images, so it can accurately identify the burr area in the image; then, the static features of the burrs are extracted from the copper rod polishing database. The static features include the shape, size, edge features, etc. of the burrs, which may have certain regularity in different burr areas; then, a sensitive convolution kernel is constructed based on the static features of the burrs extracted from the copper rod polishing database. This convolution kernel is specifically used to enhance the feature response of the burr candidate area. It pays special attention to the edge, contour and other information of the burr when processing the image, and suppresses other irrelevant areas in the image; since the burr candidate area image after binary segmentation may contain noise or unclear edges, a sensitive convolution kernel can be used to perform edge filling and noise reduction on the binary image; then a linear fitting is performed on the binary image after edge filling and noise reduction. The linear fitting process includes edge detection on the binary image after edge filling and noise reduction, determining the boundary of the burr candidate area, and then the linear fitting algorithm can be used to fit these boundary points through the least squares method to optimize the boundary of the burr area, making it smoother and consistent with the actual burr shape, thereby obtaining the optimized burr candidate area.

[0120] Embodiment 2;

[0121] Based on the same inventive concept as the copper rod electrolytic polishing burr detection method in the aforementioned embodiment, the present invention also provides a copper rod electrolytic polishing burr detection system, the system comprising:

[0122] The copper rod image acquisition module determines the moving direction of the copper rod and continuously acquires a plurality of copper rod surface images according to the moving direction of the copper rod. The plurality of copper rod surface images are multi-frame time-series images.

[0123] The electrolyte interference screening module collects the physical properties of the electrolyte, performs dynamic feature analysis on the copper rod surface images of adjacent frames based on the physical properties of the electrolyte, and identifies and screens the electrolyte interference area;

[0124] The burr candidate identification module compares and identifies the electrolyte interference area with several copper rod surface images to obtain the burr candidate area;

[0125] The burr feature screening module establishes a copper rod polishing database, extracts burr static features based on the copper rod polishing database, and screens burr candidate areas based on the burr static features to obtain burr detection results.

[0126] The above-mentioned adjustment system in the present invention can effectively realize a burr detection method for copper rod electrolytic polishing, and the technical effects that can be achieved are as described in the above-mentioned embodiments and will not be repeated here.

[0127] Specifically, the electrolyte interference screening module includes:

[0128] An image preprocessing calculation unit preprocesses the copper rod surface images of adjacent frames, performs pixel point displacement calculation on the preprocessed copper rod surface images, and obtains the original pixel displacement;

[0129] The motion compensation processing unit collects the copper rod displacement data according to the moving direction of the copper rod, and performs motion compensation on the original pixel displacement according to the copper rod displacement data to obtain the electrolyte displacement;

[0130] An interference verification marking unit verifies the electrolyte displacement according to the physical properties of the electrolyte and generates an interference marking matrix;

[0131] The interference region reconstruction unit reconstructs the topological relationship of the interference mark matrix to obtain the electrolyte interference region.

[0132] Similarly, the above-mentioned optimization schemes for the system can also respectively achieve the corresponding optimization effects of the method in Example 1, which will not be repeated here.

[0133] Although the present application has been described with reference to specific features and embodiments thereof, it is apparent that various modifications and combinations may be made thereto without departing from the spirit and scope of the present application. Accordingly, this specification and drawings are merely illustrative of the present application as defined herein and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, the present application is intended to include such modifications and variations as fall within the scope of the present application and its equivalents.

Claims

1. A method for detecting burrs during electrolytic polishing of a copper rod, characterized in that: The method comprises: Determine the moving direction of the copper rod, and continuously collect a plurality of copper rod surface images according to the moving direction of the copper rod, wherein the plurality of copper rod surface images are multi-frame time-series images; Collecting physical properties of the electrolyte, performing dynamic feature analysis on the copper rod surface images of adjacent frames based on the physical properties of the electrolyte, and identifying and screening electrolyte interference areas, including: Preprocessing the copper rod surface images of adjacent frames, performing pixel point displacement calculation on the preprocessed copper rod surface images to obtain original pixel displacement; Collecting copper rod displacement data according to the moving direction of the copper rod, and performing motion compensation on the original pixel displacement according to the copper rod displacement data to obtain electrolyte displacement; Verifying the electrolyte displacement according to the physical properties of the electrolyte to generate an interference marker matrix; Reconstructing the topological relationship of the interference marker matrix to obtain the electrolyte interference area includes: Merging adjacent interference mark points in the interference mark matrix into continuous regions to obtain an initial interference region set; Filling each initial interference region hole in the initial interference region set and smoothing the region edge to generate an optimized interference region set, wherein the initial interference region hole is a discontinuous blank region in the initial interference region; Setting an interference area threshold according to the physical properties of the electrolyte, and eliminating the optimized interference region set according to the interference area threshold; Performing liquid morphology verification on the eliminated optimized interference region set, and marking the region that meets the electrolyte flow characteristics as the electrolyte interference region; Comparing and identifying the electrolyte interference area with a plurality of copper rod surface images to obtain a burr candidate area includes: Marking and removing the electrolyte interference area according to the copper rod surface image to obtain an effective detection area; Acquire spectral imaging data of the copper rod surface image, and calculate and acquire an optical band response difference of each pixel in the effective detection area based on the spectral imaging data, wherein the optical band response difference represents a difference in reflectivity of a pixel point for two different bands; determining a difference direction according to the optical band response difference, and marking potential burr pixels according to the difference direction; Performing boundary closing on the potential burr pixels, performing physical feature verification on the burr pixels after boundary closing, and marking burr candidate areas according to the physical feature verification results; A copper rod polishing database is established, burr static features are extracted according to the copper rod polishing database, and the burr candidate areas are screened according to the burr static features to obtain a burr detection result.

2. The burr detection method for copper rod electrolytic polishing according to claim 1, characterized in that: Filling each initial interference region hole in the initial interference region set and smoothing the region edge to generate an optimized interference region set, including: identifying closed holes based on the initial interference area, wherein the closed holes represent areas completely surrounded by the initial interference area; collecting electrolysis process parameters in real time, determining the hole filling size according to the physical characteristics of the electrolyte, and determining the expanded pixel value based on the electrolysis process parameters; The hole boundary is expanded according to the hole filling size and the expanded pixel value, and the closed hole after expansion is smoothed to obtain an optimized interference area.

3. The burr detection method for copper rod electrolytic polishing according to claim 1, characterized in that: Extract burr static features, including: Acquire a historical burr detection image according to the copper rod polishing database, and perform grayscale conversion on the historical burr detection image to acquire a grayscale image; Preprocessing the grayscale image, and calculating the gradient magnitude and gradient direction based on the preprocessed grayscale image; Obtaining an angle threshold and a gradient amplitude threshold according to the copper rod polishing database, screening the angle between the gradient direction and the copper rod moving direction according to the angle threshold, obtaining a gradient response area, and dividing the gradient response area into a burr feature area and a potential burr feature area according to the gradient amplitude threshold; Performing correlation verification of the geometric shape and spectral reflectance characteristics of the potential burr feature area, and connecting with the burr feature area if a preset verification condition is met; Extract static features from the connected burr feature areas to obtain a burr static feature set.

4. The burr detection method for copper rod electrolytic polishing according to claim 3, characterized in that: Calculate the gradient magnitude and gradient direction, including: Performing convolution operations on the grayscale image in the horizontal direction and the vertical direction respectively to obtain convolution operation results; Collecting the actual surface curvature of the copper rod and the surface curvature of the image, obtaining an axial deformation compensation factor, and calculating the gradient amplitude of the pixel point according to the convolution operation result and the axial deformation compensation factor; According to the vector relationship between the convolution operation results in the horizontal and vertical directions, the gradient direction of each pixel point is calculated and obtained.

5. The burr detection method for copper rod electrolytic polishing according to claim 1, characterized in that: Marking a burr candidate area according to the physical feature verification result and optimizing the burr candidate area, including: Perform binary segmentation on the burr candidate area based on a deep learning algorithm to obtain a binary segmentation image; Extracting burr static features according to the copper rod polishing database, and constructing a sensitive convolution kernel according to the burr static features; The binary segmentation image is edge-filled and denoised according to the sensitive convolution kernel, and linear fitting is performed on the binary segmentation image after the edge-filled and denoised processing to optimize the burr candidate area.

6. A burr detection system for copper rod electrolytic polishing, characterized in that: The electropolishing burr detection method according to claim 1 is used, wherein the system comprises: A copper rod image acquisition module determines the moving direction of the copper rod and continuously acquires a plurality of copper rod surface images according to the moving direction of the copper rod, wherein the plurality of copper rod surface images are multi-frame time-series images; An electrolyte interference screening module collects physical properties of the electrolyte, performs dynamic feature analysis on the copper rod surface images of adjacent frames based on the physical properties of the electrolyte, and identifies and screens electrolyte interference areas; a burr candidate identification module, which compares and identifies the electrolyte interference area with a plurality of copper rod surface images to obtain a burr candidate area; A burr feature screening module is configured to establish a copper rod polishing database, extract burr static features based on the copper rod polishing database, and screen the burr candidate areas based on the burr static features to obtain burr detection results; The electrolyte interference screening module includes: An image preprocessing calculation unit preprocesses the copper rod surface images of adjacent frames, performs pixel point displacement calculation on the preprocessed copper rod surface images, and obtains original pixel displacement; a motion compensation processing unit, which collects copper rod displacement data according to the moving direction of the copper rod, and performs motion compensation on the original pixel displacement according to the copper rod displacement data to obtain electrolyte displacement; an interference verification marking unit, which verifies the displacement of the electrolyte according to the physical properties of the electrolyte and generates an interference marking matrix; The interference region reconstruction unit reconstructs the topological relationship of the interference mark matrix to obtain the electrolyte interference region.

Images