Burr detection method and system for copper rod electrolytic polishing

Abstract

The invention relates to the technical field of copper rod burr detection, in particular to a copper rod electrolytic polishing burr detection method and system.The method comprises the steps that firstly, surface images of a copper rod are continuously collected by determining the moving direction of the copper rod, and dynamic feature analysis is conducted on the surface images of the copper rod through collected electrolyte physical attributes; and obtaining an electrolyte interference area, carrying out comparison identification with the plurality of copper rod surface images according to the electrolyte interference area, obtaining burr candidate areas, screening the burr candidate areas according to the established copper rod polishing database, and obtaining a burr detection result. According to the invention, the problem that an efficient and stable detection result cannot be provided due to the fact that burr detection is easily interfered by human factors is effectively solved, the precision and reliability of burr detection are improved, and the detection efficiency and the product quality of copper rod electrolytic polishing are remarkably improved.

Description

Technical Field

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

[0002] With the rapid development of modern manufacturing and the increasing demand for precision machining, the electrolytic polishing technology of copper rods has been widely used in the fields of electricity, electronics, etc. Electrolytic polishing can not only improve the surface finish of copper rods, but also effectively improve their corrosion resistance and electrical conductivity, so it is crucial in the production process of many high-precision products. However, during the electrolytic polishing process, burrs are likely to appear on the surface of copper rods, affecting subsequent processing techniques and product quality. The presence of burrs may not only affect the appearance quality of copper rods, but also lead to unstable electrical conductivity or poor connection, thus affecting the reliability and performance of the entire electronic product.

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

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

[0005] The present invention provides a burr detection method and system for electrolytic polishing of copper rods, which can effectively solve the problems in the background art.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is: A burr detection method for electrolytic polishing of copper rods, the method comprising: 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, and the plurality of copper rod surface images are multiple frames of sequential images; Collect the physical properties of the electrolyte, and perform dynamic feature analysis on the copper rod surface images of adjacent frames according to the physical properties of the electrolyte to identify and screen the electrolyte interference regions; Compare and identify the electrolyte interference regions with the plurality of copper rod surface images respectively to obtain burr candidate regions; Establish a copper rod polishing database, extract the static burr features according to the copper rod polishing database, and screen the burr candidate areas according to the static burr features to obtain the burr detection results.

[0007] Further, perform dynamic feature analysis on the copper rod surface images of adjacent frames according to the physical properties of the electrolyte, and identify and screen the electrolyte interference areas, including: Preprocess the copper rod surface images of adjacent frames, calculate the pixel point displacement of the preprocessed copper rod surface images to obtain the original pixel displacement; Collect the copper rod displacement data according to the copper rod moving direction, and perform motion compensation on the original pixel displacement according to the copper rod displacement data to obtain the electrolyte displacement; Verify the electrolyte displacement according to the physical properties of the electrolyte to generate an interference marking matrix; Reconstruct the topological relationship of the interference marking matrix to obtain the electrolyte interference area.

[0008] Further, reconstruct the topological relationship of the interference marking matrix to obtain the electrolyte interference area, including: Merge adjacent interference marking points in the interference marking matrix into continuous areas to obtain an initial interference area set; Fill the holes of each initial interference area in the initial interference area set, and smooth the area edges to generate an optimized interference area set, where the initial interference area holes are non - continuous blank areas within the initial interference areas; Set an interference area threshold according to the physical properties of the electrolyte, and eliminate the optimized interference area set according to the interference area threshold; Verify the liquid form of the optimized interference area set after elimination, and mark the areas that conform to the electrolyte flow characteristics as electrolyte interference areas.

[0009] Further, fill the holes of each initial interference area in the initial interference area set, and smooth the area edges to generate an optimized interference area set, including: Identify the closed holes according to the initial interference areas, where the closed holes represent areas completely surrounded by the initial interference areas; Real - time collect the electrolysis process parameters, determine the hole filling size according to the physical characteristics of the electrolyte, and determine the extended pixel value based on the electrolysis process parameters; Expand the hole boundaries according to the hole filling size and the extended pixel value, and smooth the expanded closed holes to obtain the optimized interference areas.

[0010] Further, by comparing and identifying the electrolyte interference regions with a plurality of the surface images of the copper rods respectively, a burr candidate region is obtained, including: Mark and remove the electrolyte interference regions according to the surface images of the copper rods to obtain an effective detection region; Obtain spectral imaging data of the surface images of the copper rods, and calculate and obtain the optical band response difference of each pixel in the effective detection region according to the spectral imaging data, where the optical band response difference represents the difference in reflectivity of a pixel point for two different bands; Determine the difference direction according to the optical band response difference, and mark potential burr pixels according to the difference direction; Close the boundaries of the potential burr pixels, verify the physical characteristics according to the burr pixels after boundary closing, and mark the burr candidate region according to the physical characteristic verification result.

[0011] Further, static burr features are extracted, including: Obtain historical burr detection images according to the copper rod polishing database, convert the historical burr detection images to grayscale to obtain grayscale images; Preprocess the grayscale images, and calculate the gradient magnitude and gradient direction according to the preprocessed grayscale images; Obtain an angle threshold and a gradient magnitude threshold according to the copper rod polishing database, screen the angle between the gradient direction and the moving direction of the copper rod according to the angle threshold to obtain a gradient response region, and divide the gradient response region into a burr feature region and a potential burr feature region according to the gradient magnitude threshold; Perform an association verification on the geometric shape and spectral reflection characteristics of the potential burr feature region, and if the preset verification conditions are met, connect it to the burr feature region; Extract static features from the connected burr feature region to obtain a set of static burr features.

[0012] Further, the gradient magnitude and gradient direction are calculated, including: Perform convolution operations on the grayscale images based on the horizontal direction and the vertical direction respectively to obtain the convolution operation results; Collect the actual surface curvature and the image surface curvature of the copper rod to obtain an axial deformation compensation factor, and calculate the gradient magnitude of the pixel points according to the convolution operation results and the axial deformation compensation factor; Calculate and obtain the gradient direction of each pixel point according to the vector relationship between the horizontal and vertical directions of the convolution operation results.

[0013] Further, mark the burr candidate region according to the physical characteristic verification result, and optimize the burr candidate region, including: Perform binary segmentation on the burr candidate area based on a deep learning algorithm to obtain a binary segmentation image; Extract burr static features according to the copper rod polishing database, and construct a sensitive convolutional kernel according to the burr static features; Perform edge filling and noise reduction processing on the binary segmentation image according to the sensitive convolutional kernel, perform linear fitting on the binary segmentation image after the edge filling and noise reduction processing, and optimize the burr candidate area.

[0014] A burr detection system for electrolytic polishing of copper rods, the system includes: A copper rod image acquisition module, which 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; An electrolyte interference screening module, which collects the physical properties of the electrolyte, performs dynamic feature analysis on the copper rod surface images of adjacent frames according to the physical properties of the electrolyte, and identifies and screens the electrolyte interference area; A burr candidate recognition module, which performs comparative recognition on the electrolyte interference area and a plurality of the copper rod surface images respectively to obtain a burr candidate area; A burr feature screening module, which establishes a copper rod polishing database, extracts burr static features according to the copper rod polishing database, and screens the burr candidate area according to the burr static features to obtain a burr detection result.

[0015] Further, the electrolyte interference screening module includes: An image preprocessing calculation unit, which 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; 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 the electrolyte displacement; An interference verification marking unit, which verifies the electrolyte displacement according to the physical properties of the electrolyte and generates an interference marking matrix; An interference area reconstruction unit, which reconstructs the topological relationship of the interference marking matrix to obtain the electrolyte interference area.

[0016] Through the technical solution of the present invention, the following technical effects can be achieved: Effectively solves the problem that it is difficult to identify due to the high similarity in optical characteristics between the dynamic interference caused by the electrolyte flow and the real burrs. Through the 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 result, improve the accuracy of burr detection, improve the surface quality of the copper rod, optimize the production process, and reduce the defective rate.

[0017] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of this application more obvious and understandable, the specific embodiments of this application are specifically listed below. Brief Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0019] Figure 1 It is a schematic flow chart of the burr detection method for copper rod electrolytic polishing; Figure 2 It is a schematic flow chart of the identification of the electrolyte interference area; Figure 3 It is a schematic flow chart of the topological reconstruction of the interference marking matrix; Figure 4 It is a schematic diagram of the topological reconstruction of the interference marking matrix. Detailed Embodiments

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this invention belongs. The terms used in the specification of this invention are only for the purpose of describing specific embodiments, and are not intended to limit this invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0022] Embodiment 1; As Figure 1 shown, this application provides a burr detection method for copper rod electrolytic polishing, and the method includes: S10: Determine the moving direction of the copper rod, and continuously collect a number of copper rod surface images according to the moving direction of the copper rod. The number of copper rod surface images is multiple frames of sequential images; S20: Collect the physical properties of the electrolyte, perform dynamic feature analysis on the copper rod surface images of adjacent frames according to the physical properties of the electrolyte, and identify and screen the electrolyte interference areas; S30: Compare and identify the electrolyte interference areas with a number of copper rod surface images respectively to obtain the burr candidate areas; S40: Establish a copper rod polishing database, extract the static features of the burrs according to the copper rod polishing database, and screen the burr candidate areas according to the static features of the burrs to obtain the burr detection results.

[0023] 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. By precisely measuring the rotational or linear displacement of the copper rod, the encoder can provide accurate synchronization information for subsequent image acquisition. Alternatively, a photoelectric sensor installed on the polishing machine can be used to detect the displacement of the copper rod to determine its movement direction. According to the determined moving direction of the copper rod, continuously acquire several surface images of the copper rod. These images are multi-frame sequential images, and each frame corresponds to a time point. During the image acquisition process, a high-speed industrial camera can be used to capture the images, and the acquisition frequency should be adjusted according to the moving speed of the copper rod. To ensure the integrity and details of the image data, a high-resolution camera can be used. At the same time as the image acquisition, the physical properties of the electrolyte (such as flow rate, temperature, concentration, and viscosity, etc.) need to be acquired in real time. The physical properties of the electrolyte have an important impact on the copper rod polishing process. In particular, the flow rate and temperature directly affect the flow of the electrolyte and the polishing effect on the surface of the copper rod. Therefore, dedicated sensors for the physical properties of the electrolyte, such as flow meters, temperature sensors, and concentration sensors, are used to ensure real-time data acquisition. After acquiring the physical properties of the electrolyte, then perform dynamic feature analysis on adjacent frames of the copper rod surface images. Dynamic feature analysis refers to extracting features related to time changes by analyzing the changes between consecutive frames to facilitate the accurate identification of dynamic elements in the images. In the present invention, the purpose of dynamic feature analysis is to capture and analyze the interference regions caused by electrolyte flow, bubbles, or foams, etc. from adjacent frames of the copper rod surface images. In this way, the difference between these interference factors and the true features of the copper rod surface can be effectively distinguished, thereby identifying and screening the electrolyte interference regions. Then, compare and identify the electrolyte interference regions with the copper rod surface images to obtain the burr candidate regions. To improve the accuracy of burr detection, this method uses 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 the size, shape, edge smoothness, etc. of the burrs. Then, screen the burr candidate regions according to the burr static features to finally obtain the burr detection result.

[0024] Through the technical solution of the present invention, the problem that it is difficult to identify due to the high similarity in optical features between the dynamic interference caused by electrolyte flow and real burrs is effectively solved. Through the comprehensive analysis of the copper rod surface images and the physical properties of the electrolyte, the electrolyte interference regions can be accurately identified and screened, effectively avoiding the influence of electrolyte interference on the burr detection result, improving the accuracy of burr detection, enhancing the surface quality of the copper rod, optimizing the production process, and reducing the defective product rate.

[0025] Furthermore, as Figure 2As shown, perform dynamic feature analysis on the surface images of copper rods in adjacent frames according to the physical properties of the electrolyte, and identify and screen the electrolyte interference regions, including: S21: Preprocess the surface images of copper rods in adjacent frames, calculate the pixel point displacement of the preprocessed surface images of copper rods, and obtain the original pixel displacement; S22: Collect copper rod displacement data according to the copper rod movement direction, and perform motion compensation on the original pixel displacement according to the copper rod displacement data to obtain the electrolyte displacement; S23: Verify the electrolyte displacement according to the physical properties of the electrolyte, and generate an interference marking matrix; S24: Reconstruct the topological relationship of the interference marking matrix to obtain the electrolyte interference region.

[0026] As an optimization of the above embodiments, first, preprocess the acquired surface image of the copper rod. The preprocessing includes steps such as denoising, image sharpening, and contrast enhancement to improve the image quality and provide clearer image data for subsequent pixel displacement calculation. Since there will be a certain movement or displacement on the surface of the copper rod over time in continuously acquired adjacent frame images, which causes errors in the actual measurement of the electrolyte displacement, it is necessary to calculate the pixel point displacement of the preprocessed surface image of the copper rod as follows: First, register adjacent frames, which can be achieved through feature-based image registration algorithms (such as SIFT, SURF, or ORB algorithms) to find the matching points between the images. Then, calculate the pixel displacement through the optical flow method as follows: Assume that the brightness of the object is constant between frames with a very short time interval. Then, by analyzing the brightness changes of each pixel in adjacent frames and comparing the differences in the brightness of each pixel point between the two frames, the displacement of the pixel point is deduced to obtain the original pixel displacement. The movement of the pixel point can be represented by a velocity vector, which describes the position change of the pixel from the previous frame to the current frame. And because 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 adjusts 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 result and finally obtain the electrolyte displacement. Because physical properties of the electrolyte, such as flow rate, temperature, etc., will affect its flow on the surface of the copper rod, it is necessary to verify the electrolyte displacement. In some embodiments, the fast matching method can be used. Divide the surface image of the copper rod into multiple small blocks and compare them with the corresponding regions in the next time frame. By calculating the similarity between these small blocks, the system can infer the displacement amount of each small block. Especially on the surface of the copper rod, the electrolyte flow will cause changes in the pixels in some areas. The system can accurately track these changes using the block matching method and identify and mark the interference regions caused by electrolyte flow or bubbles. Then, by comparing the displacement changes brought about by the electrolyte flow, the system can identify the interference regions in the image caused by factors such as electrolyte flow, bubbles, or foam. These interference regions usually manifest as local discontinuities or irregularities in the image, especially in areas with fast or unstable flow, which may lead to relatively large displacements between pixels. After identifying these regions, the system generates an interference marking matrix, where each element of the matrix represents whether the corresponding position in the image is an interference region.If a pixel position is affected by electrolyte flow or bubbles, its corresponding matrix element will be marked as interference (e.g., marked as 1); if not affected, the element is marked as non-interference (e.g., marked as 0). The generated interference marking matrix can be used to screen out the areas in the image affected by factors such as electrolyte flow and bubbles. These areas are usually excluded in subsequent steps to avoid unnecessary interference with burr detection. Finally, the topological relationship of the interference marking matrix is reconstructed to obtain the electrolyte interference area.

[0027] Furthermore, as Figure 3 shown in Figure 4 , reconstructing the topological relationship of the interference marking matrix to obtain the electrolyte interference area includes: S241: Merge adjacent interference marking points in the interference marking matrix into continuous regions to obtain an initial set of interference regions; S242: Fill the holes of each initial interference region in the initial set of interference regions and smooth the region edges to generate an optimized set of interference regions. The initial interference region holes are non-continuous blank regions within the initial interference regions; S243: Set an interference area threshold according to the physical properties of the electrolyte and exclude the optimized set of interference regions according to the interference area threshold; S244: Verify the liquid morphology of the optimized set of interference regions after exclusion, and mark the regions that conform to the electrolyte flow characteristics as electrolyte interference areas.

[0028] In this embodiment, first, the system traverses each pixel point in the interference marking matrix. For each interference marking point (e.g., marked as "1"), the system checks the pixel points in its eight neighborhoods. If any pixel point in these neighborhoods is also marked as "1", the system considers these two pixel points to be connected and they belong to the same interference region. At this time, the system merges these two pixel points, and the merged marking point will be given the same region number or label, forming a connected region, and obtaining the initial set of interference regions. Next, each region in the initial set of interference regions is further processed. It is checked whether there are holes (i.e., non - continuous blank regions within the region) in each initial interference region. If there are holes, the "hole filling" technique in image processing is usually used to fill these blank regions; and the edges of each region are smoothed to remove the jagged edges caused by electrolyte interference or image noise, thereby generating an optimized set of interference regions. After that, further screening is performed based on the optimized set of interference regions according to the physical properties of the electrolyte. A threshold for the interference area is set and judged according to physical characteristics such as the flow rate and temperature of the electrolyte. If the area of a certain interference region is less than the set threshold, this region will be considered an invalid region and removed from the optimized set of interference regions; through this screening step, the system can effectively remove smaller interference regions that may be caused by bubbles or minor perturbations, reducing false detections; after removing the invalid regions, further liquid morphology verification is performed on the optimized set of interference regions. Liquid morphology verification refers to detecting and verifying the morphological characteristics of the electrolyte interference regions to determine whether these regions conform to the natural characteristics of electrolyte flow. The purpose is to ensure that the identified interference regions are caused by factors such as electrolyte flow, bubbles, or foam, rather than other noise or mis - marked interference regions; the method of liquid morphology verification can refer to the following method: morphological analysis (such as erosion or dilation) is performed according to the electrolyte flow characteristics (such as fluidity, morphology, etc.), and this analysis considers the movement trajectory of the electrolyte on the copper rod surface to ensure that the detected interference regions conform to the actual characteristics of electrolyte flow. The regions that conform to the flow characteristics will be marked as electrolyte interference regions.

[0029] Furthermore, filling the holes in each initial interference region in the initial set of interference regions and smoothing the region edges to generate an optimized set of interference regions includes: Identifying closed holes according to the initial interference region. A closed hole represents a region completely surrounded by the initial interference region; Real - time collecting electrolysis process parameters, determining the hole filling size according to the physical characteristics of the electrolyte, and determining the extended pixel value based on the electrolysis process parameters; Expanding the hole boundary according to the hole filling size and the extended pixel value, and smoothing the expanded closed hole to obtain an optimized interference region.

[0030] Specifically, in the initial set of interference regions, image processing methods (such as erosion and dilation in morphological operations) can be used to identify enclosed holes. Enclosed holes refer to those blank regions that are completely surrounded by the initial interference regions, which are usually caused by uneven electrolyte flow, bubbles, or other external factors. After that, electrolysis process parameters are collected in real time, including electrolyte flow rate, temperature, and current density, etc., which have a direct impact on the flow characteristics of the electrolyte on the surface of the copper rod. Then, according to 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: Calculate the area of the hole based on the pixel values of the region. The filling size is proportional to the area of the hole, and the fluidity of the electrolyte also affects the filling size. An 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 (low viscosity), a larger filling size will be used to simulate the rapid filling of the electrolyte flow, and the extended pixel value is calculated according to the electrolysis process parameters (including electrolyte fluidity, electrolyte temperature, and current density of the electrolyte, etc.). The calculation formula is as follows: ; Wherein, represents the extended pixel value, indicating the pixel size of the hole expansion, represents the electrolyte fluidity, represents the electrolyte temperature, represents the standard temperature, is the current density of the electrolyte, represents the standard current density, represents the adjustment coefficient, obtained based on historical experience. Then, the hole boundary is expanded according to the hole filling size and the extended pixel value, and a smoothing algorithm (such as Gaussian filtering) is applied to process the expanded hole boundary to make it blend with the surrounding region edges, obtaining an optimized interference region.

[0031] Furthermore, by comparing and identifying the electrolyte interference regions with several copper rod surface images respectively, a burr candidate region is obtained, including: Mark and remove the electrolyte interference regions according to the copper rod surface image to obtain an effective detection region; Obtain the 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 region according to the spectral imaging data. The optical band response difference represents the difference in reflectance of a pixel point for two different bands; Determine the difference direction according to the optical band response difference, and mark potential burr pixels according to the difference direction; Perform boundary closure on potential burr pixels, verify physical features based on the burr pixels after boundary closure, and mark burr candidate regions according to the results of physical feature verification.

[0032] As an optimization of the above embodiment, first remove the electrolyte interference region 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 of each pixel; multi-band reflectance refers to the light reflection ability of the object surface at multiple different wavelengths (or bands); different substances have different reflection characteristics for light in different bands. Therefore, by measuring the reflectance at multiple bands, rich information about the composition, structure, and characteristics of the object surface can be obtained. Then, by comparing the reflectances between different spectral bands, calculate the difference in band reflectance of each pixel point to obtain the optical band response difference. This difference represents the reflectance difference of this pixel point at different bands, and this difference can be used as an effective feature for identifying burrs because the optical reflection characteristics between burrs and the surrounding surface are quite different, so it can significantly highlight the difference in response at different bands; analyze the direction of these differences based on the calculated optical band response differences. The optical difference direction usually refers to in the spectral data of multiple bands, in which direction or which band's reflectance change the reflectance difference of a specific pixel point is mainly manifested. In short, the optical difference direction describes the directionality of the reflectance difference between different spectral bands, indicating which band contributes the most to the reflectance difference of this pixel among multiple spectral bands; by calculating the response difference of each band, find the direction with the largest difference or the most obvious difference in the entire image. This direction can be used as a discriminant basis for regions such as burrs and stains, thereby obtaining potential burr pixels; then morphological operations such as erosion and dilation can be used to eliminate breaks or gaps caused by image noise or interference, fill the gaps in the burr region, and thus form a complete burr candidate region.

[0033] Furthermore, extract static burr features, including: Obtain historical burr detection images from the copper rod polishing database, perform grayscale conversion on the historical burr detection images to obtain grayscale images; Preprocess the grayscale images and calculate the gradient magnitude and gradient direction based on the preprocessed grayscale images; Obtain the angle threshold and gradient magnitude threshold from the copper rod polishing database, screen the angle between the gradient direction and the copper rod movement direction according to the angle threshold to obtain the gradient response region, and divide the gradient response region into a burr feature region and a potential burr feature region according to the gradient magnitude threshold; Perform correlation verification on the geometric morphology and spectral reflection characteristics of the potential burr feature region. If the preset verification conditions are met, connect it to the burr feature region; Extract static features from the burr feature area after connection to obtain the burr static feature set.

[0034] In this embodiment, first, historical burr detection images are obtained from the copper rod polishing database. The RGB channel values of each pixel in the historical burr detection images are extracted, and the RGB channel values are weighted and averaged to obtain the grayscale values of several pixels. The grayscale values are integrated to generate a grayscale image. To improve the accuracy of subsequent detection, the system preprocesses the grayscale image. The preprocessing steps usually include noise removal, image enhancement, contrast adjustment, etc., in order to highlight the details of the burr area. After preprocessing, the gradient magnitude and gradient direction of the image are calculated based on the pixel values of the image. The gradient magnitude is the intensity of the change in pixel values in the image, reflecting the speed of surface change, while the gradient direction represents the direction of the change in pixel values. After calculating and obtaining the gradient magnitude and gradient direction, an included angle threshold is set. For example, a range with an included angle of ±15° or ±30° is set. If the included angle between the gradient direction of a certain pixel point and the moving direction of the copper rod falls within this range, it is considered that this area is aligned with the moving direction of the copper rod and there may be burrs. Only when the included angle between the gradient direction and the moving direction of the copper rod is less than the preset included angle threshold, will this pixel point be regarded as having a strong response, indicating that there may be burrs in this area. The gradient magnitude reflects the intensity of the brightness change within the area. On the surface of the copper rod, the presence of burrs usually leads to a drastic change in surface brightness, thus showing a relatively large gradient magnitude. The system will set a minimum gradient magnitude threshold according to the brightness distribution of the copper rod surface image. For example, a gradient magnitude threshold of 5 or 10 is set, which depends on the resolution of the image and the desired response intensity. Combining the results of gradient direction screening and gradient magnitude screening, the system will mark those pixel areas that simultaneously meet the conditions that the gradient direction is aligned with the moving direction of the copper rod and the gradient magnitude is greater than the threshold as potential burr feature areas. Then, the correlation verification of the geometric morphology and spectral reflection characteristics of the potential burr areas is carried out. The purpose of geometric morphology verification is to determine whether the shape characteristics of the potential burr areas conform to the common geometric characteristics of burrs. Burrs usually have specific geometric morphologies, such as sharp edges, protruding structures, or irregular shapes. In some embodiments, geometric morphology verification is usually carried out by morphological analysis. The purpose of spectral reflection characteristic verification is to confirm whether this area is a burr by analyzing the reflection characteristics of the potential burr area through spectral imaging data, and calculate the reflection differences of these areas in different spectral bands to confirm the nature of the burr. Combining the results of geometric morphology verification and spectral reflection characteristics results, the authenticity of the potential burr area is comprehensively judged. If the geometric morphology (such as sharp edges, protruding morphology) of a certain potential burr area and its reflection characteristics in multiple spectral bands conform to the preset standards, then this area is considered to be an actual burr area. After the correlation verification of the geometric morphology and spectral reflection characteristics of the potential burr areas, they are connected to the burr feature areas. These connected burr feature areas will further extract static features. The process of static feature extraction includes calculating various geometric features such as the area, shape, and edge smoothness of the burr area, so as to obtain a set of burr static features.

[0035] Furthermore, calculating the gradient magnitude and gradient direction includes: Performing convolution operations on the grayscale image separately in the horizontal and vertical directions to obtain the results of the convolution operations; Collecting the actual surface curvature and the image surface curvature of the copper rod to obtain the axial deformation compensation factor, and calculating the gradient magnitude of the pixel points according to the results of the convolution operations and the axial deformation compensation factor; Calculating and obtaining the gradient direction of each pixel point according to the vector relationship of the results of the convolution operations in the horizontal and vertical directions.

[0036] Specifically, first, the Sobel operator can be used to perform convolution on the image in the horizontal and vertical directions respectively. The convolution operations in these two directions can extract the features of the horizontal edges and vertical edges in the image respectively, so as to obtain the local gradient information of each pixel point in the horizontal and vertical directions. The results of the convolution operations will provide a horizontal and vertical gradient value for each pixel point in the image; then, the values in the horizontal and vertical directions of the results of the convolution operations are combined to calculate the original gradient magnitude of each pixel point. The original gradient magnitude represents the intensity of the brightness change at a certain point in the image, and its calculation formula is: ; where, and are the gradient values in the horizontal and vertical directions; since the surface of the copper rod may be affected by physical deformations, such as surface bending or irregularities, in order to eliminate the influence of these deformations on the gradient calculation, compensation will also be carried out by collecting the actual surface curvature data of the copper rod: using a surface curvature sensor or an image analysis algorithm to measure the degree of bending of the copper rod surface, and calculating the deformation compensation factor according to the surface curvature data, and adjusting the original gradient magnitude according to the compensation factor to eliminate the error caused by surface bending and ensure the accuracy of the gradient magnitude and gradient direction; then, according to the gradient values in the horizontal and vertical directions, calculating the gradient direction of each pixel point, and the calculation formula is: ; where, and are the gradient values in the horizontal and vertical directions.

[0037] Furthermore, marking the burr candidate areas according to the physical feature verification results and optimizing the burr candidate areas includes: Performing binary segmentation on the burr candidate areas based on a deep learning algorithm to obtain a binary segmentation image; Extracting the static features of the burrs according to the copper rod polishing database and constructing a sensitive convolution kernel according to the static features of the burrs; Perform edge filling and noise reduction on the binary segmentation image according to the sensitive convolution kernel, perform linear fitting on the binary segmentation image after edge filling and noise reduction, and optimize the burr candidate area.

[0038] As a preference of the above embodiment, first, a deep learning algorithm is 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. This 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, extract the static features of the burrs from the copper rod polishing database. The static features include the shape, size, edge features, etc. of the burrs, and they may have certain regularities in different burr areas. Then, according to the static features of the burrs extracted from the copper rod polishing database, construct a sensitive convolution kernel. This convolution kernel is specifically used to enhance the feature response of the burr candidate area. It will pay special attention to the edges, contours, etc. of the burrs during image processing and suppress other irrelevant areas in the image. Since the burr candidate area image after binary segmentation may contain noise or unclear edges, for this reason, a sensitive convolution kernel can be used to perform edge filling and noise reduction on the binary image. Subsequently, perform linear fitting on the binary image after edge filling and noise reduction. The process of linear fitting includes performing edge detection on the binary image after edge filling and noise reduction to determine the boundary of the burr candidate area, and then the linear fitting algorithm can be used to fit these boundary points by the least squares method, so as to optimize the boundary of the burr area, make it smoother and conform to the actual burr shape, and then obtain the optimized burr candidate area.

[0039] Embodiment 2; Based on the same inventive concept as a burr detection method for copper rod electrolytic polishing in the foregoing embodiment, the present invention also provides a burr detection system for copper rod electrolytic polishing. The system includes: A copper rod image acquisition module, which 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; An electrolyte interference screening module, which acquires the physical properties of the electrolyte and performs dynamic feature analysis on the copper rod surface images of adjacent frames according to the physical properties of the electrolyte to identify and screen the electrolyte interference area; A burr candidate recognition module, which respectively compares and recognizes the electrolyte interference area with a plurality of copper rod surface images to obtain a burr candidate area; A burr feature screening module, which establishes a copper rod polishing database, extracts the static features of the burrs according to the copper rod polishing database, and screens the burr candidate area according to the static features of the burrs to obtain the burr detection result.

[0040] The above adjustment system in the present invention can effectively implement a burr detection method for electrolytic polishing of copper rods, and the technical effects that can be achieved are as described in the above embodiments, which will not be elaborated here.

[0041] Furthermore, the electrolyte interference screening module includes: An image preprocessing calculation unit preprocesses the copper rod surface images of adjacent frames, calculates the pixel point displacement of the preprocessed copper rod surface images, and obtains the original pixel displacement; A motion compensation processing unit collects copper rod displacement data according to the copper rod movement direction, and performs motion compensation on the original pixel displacement according to the copper rod displacement data to obtain the electrolyte displacement; An interference verification marking unit verifies the electrolyte displacement according to the physical properties of the electrolyte to generate an interference marking matrix; An interference region reconstruction unit reconstructs the topological relationship of the interference marking matrix to obtain the electrolyte interference region.

[0042] Similarly, for the above optimization solutions of the system, the corresponding optimization effects of the methods in Embodiment 1 can also be respectively achieved, which will not be elaborated here either.

[0043] Although the present application has been described in combination with specific features and their embodiments, it is obvious that various modifications and combinations can be made without departing from the spirit and scope of the present application. Accordingly, the present specification and the drawings are only exemplary descriptions of the present application defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. A burr detection method for electrolytic polishing of copper rods, characterized in that, The method includes: 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. The plurality of copper rod surface images are multiple frames of sequential images; Collect the physical properties of the electrolyte, and perform dynamic feature analysis on the copper rod surface images of adjacent frames according to the physical properties of the electrolyte to identify and screen the electrolyte interference regions; Compare and identify the electrolyte interference regions with the plurality of copper rod surface images respectively to obtain burr candidate regions; Establish a copper rod polishing database, extract the static features of the burrs according to the copper rod polishing database, and screen the burr candidate regions according to the static features of the burrs to obtain the burr detection results.

2. The burr detection method for electrolytic polishing of copper rods according to claim 1, characterized in that, Performing dynamic feature analysis on the copper rod surface images of adjacent frames according to the physical properties of the electrolyte to identify and screen the electrolyte interference regions, including: Preprocess the copper rod surface images of adjacent frames, calculate the pixel point displacement of the preprocessed copper rod surface images to obtain the original pixel displacement; Collect the copper rod displacement data according to the moving direction of the copper rod, and perform motion compensation on the original pixel displacement according to the copper rod displacement data to obtain the electrolyte displacement; Verify the electrolyte displacement according to the physical properties of the electrolyte to generate an interference marking matrix; Reconstruct the topological relationship of the interference marking matrix to obtain the electrolyte interference region.

3. The burr detection method for electrolytic polishing of copper rods according to claim 2, wherein, Reconstruct the topological relationship of the interference marking matrix to obtain the electrolyte interference region, including: Merge adjacent interference marking points in the interference marking matrix into continuous regions to obtain an initial interference region set; Fill the holes of each initial interference region in the initial interference region set and smooth the region edges to generate an optimized interference region set. The initial interference region holes are non - continuous blank regions within the initial interference regions; Set an interference area threshold according to the physical properties of the electrolyte, and eliminate the optimized interference region set according to the interference area threshold; Verify the liquid morphology of the optimized interference region set after elimination, and mark the regions that conform to the electrolyte flow characteristics as electrolyte interference regions.

4. The burr detection method for electrolytic polishing of copper rods according to claim 3, characterized in that, Filling the holes of each initial interference region in the initial interference region set and smoothing the region edges to generate an optimized interference region set, including: Identify the closed holes according to the initial interference regions. The closed holes represent regions completely surrounded by the initial interference regions; Collect the electrolysis process parameters in real - time, determine the hole filling size according to the physical characteristics of the electrolyte, and determine the extended pixel value based on the electrolysis process parameters; Expand the hole boundaries according to the hole filling size and the extended pixel value, and smooth the expanded closed holes to obtain the optimized interference regions.

5. The burr detection method for electrolytic polishing of copper rods according to claim 1, characterized in that, Compare and identify the electrolyte interference regions with the plurality of copper rod surface images respectively to obtain burr candidate regions, including: Mark and remove the electrolyte interference regions according to the copper rod surface images to obtain an effective detection region; Obtain the spectral imaging data of the copper rod surface image, calculate and obtain the optical band response difference of each pixel in the effective detection area according to the spectral imaging data, where the optical band response difference represents the difference in reflectivity of a pixel point for two different bands; Determine the difference direction according to the optical band response difference, and mark potential burr pixels according to the difference direction; Perform boundary closure on the potential burr pixels, verify the physical characteristics according to the burr pixels after boundary closure, and mark the burr candidate area according to the physical characteristic verification result.

6. The burr detection method for electrolytic polishing of copper rods according to claim 1, characterized in that, Extract burr static features, including: Obtain historical burr detection images according to the copper rod polishing database, perform gray conversion on the historical burr detection images to obtain gray images; Preprocess the gray image, and calculate the gradient magnitude and gradient direction according to the preprocessed gray image; Obtain the included angle threshold and gradient magnitude threshold according to the copper rod polishing database, screen the included angle between the gradient direction and the copper rod movement direction according to the included angle threshold to obtain the gradient response area, and divide the gradient response area into a burr feature area and a potential burr feature area according to the gradient magnitude threshold; Perform correlation verification on the geometric shape and spectral reflection characteristics of the potential burr feature area, and if the preset verification conditions are met, connect it to the burr feature area; Extract static features from the connected burr feature area to obtain a burr static feature set.

7. The burr detection method for electrolytic polishing of copper rods according to claim 6, characterized in that, Calculate the gradient magnitude and gradient direction, including: Perform convolution operations on the gray image based on the horizontal direction and the vertical direction respectively to obtain the convolution operation results; Collect the actual surface curvature and image surface curvature of the copper rod, obtain the axial deformation compensation factor, and calculate the gradient magnitude of the pixel points according to the convolution operation results and the axial deformation compensation factor; Calculate and obtain the gradient direction of each pixel point according to the vector relationship between the horizontal and vertical directions of the convolution operation results.

8. The burr detection method for electrolytic polishing of copper rods according to claim 5, characterized in that, Mark the burr candidate area according to the physical characteristic verification result, and optimize 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; Extract burr static features according to the copper rod polishing database, and construct a sensitive convolution kernel according to the burr static features; Perform edge filling and noise reduction processing on the binary segmentation image according to the sensitive convolution kernel, and perform linear fitting on the binary segmentation image after the edge filling and noise reduction processing to optimize the burr candidate area.

9. A burr detection system for electrolytic polishing of copper rods, characterized in that, The system includes: A copper rod image acquisition module, which determines the copper rod movement direction and continuously acquires a plurality of copper rod surface images according to the copper rod movement direction. The plurality of copper rod surface images are multi-frame time series images; An electrolyte interference screening module, which collects the physical properties of the electrolyte, performs dynamic feature analysis on the copper rod surface images of adjacent frames according to the physical properties of the electrolyte, and identifies and screens the electrolyte interference area; A burr candidate recognition module, which performs comparison and recognition on the electrolyte interference area and a plurality of copper rod surface images respectively to obtain the burr candidate area; The burr feature screening module establishes a copper rod polishing database, extracts the static burr features according to the copper rod polishing database, and screens the burr candidate areas according to the static burr features to obtain the burr detection results.

10. The burr detection system for electrolytic polishing of copper rods according to claim 9, characterized in that, The electrolyte interference screening module includes: An image preprocessing calculation unit preprocesses the copper rod surface images of adjacent frames, calculates the pixel point displacement of the preprocessed copper rod surface images, and obtains the original pixel displacement; A motion compensation processing unit collects copper rod displacement data according to the copper rod movement direction, and performs motion compensation on the original pixel displacement according to the copper rod displacement data to obtain the electrolyte displacement; An interference verification marking unit verifies the electrolyte displacement according to the physical properties of the electrolyte to generate an interference marking matrix; An interference area reconstruction unit reconstructs the topological relationship of the interference marking matrix to obtain the electrolyte interference area.

Images