A method and system for detecting edge burrs on composite panels

Abstract

This invention belongs to the field of image processing technology, specifically relating to a method and system for detecting edge burrs on composite plates. The method includes: acquiring an image of the composite plate and extracting edge baselines; obtaining a first enhancement image and a second enhancement image through morphological operations and positional subtraction; determining near-edge regions based on the distance between candidate regions and the edge baselines, and constructing a refined image based on the anisotropy index of the near-edge regions; extracting pixel sequences along the normal direction on the refined image using edge baseline pixels as target pixels to calculate the total energy, and obtaining a total energy sequence; merging the total energy sequences into sequence segments, and determining burr defects based on the defect index of the sequence segments. This invention utilizes the morphological scale difference between burrs and background textures and the extension characteristics of burrs to improve the accuracy of burr detection.

Description

Technical Field

[0001] This invention relates to the field of image processing technology. More specifically, this invention relates to a method and system for detecting burrs on the edges of composite plates. Background Technology

[0002] During the production and processing of composite panels, burrs are easily generated on their edges, affecting product quality and subsequent use. Therefore, accurate detection of burrs on the edges of composite panels is a crucial aspect of quality control.

[0003] In related technologies, for example, Chinese patent application CN118570072A discloses a method for burr detection during titanium metal processing. This method includes: first, acquiring a grayscale image of the titanium metal surface to be processed, and performing edge detection to obtain processing edge lines, suspected burr edge lines, and texture edge lines in the grayscale image; based on the grayscale changes of each pixel on the suspected burr edge line and the difference in grayscale changes compared to the pixels on the processing edge line, obtaining the burr probability of the suspected burr edge line; analyzing the gradient values ​​of each pixel on the reference texture edge line of the suspected burr edge line to obtain the burr blur factor of the suspected burr edge line; then, filtering each suspected burr edge line based on the obtained noise influence factor to obtain an enhanced image; and performing edge detection on the enhanced image to obtain the burr locations in the enhanced image.

[0004] However, related technologies primarily focus on addressing pixel grayscale variations caused by random noise, and their technical logic relies on complex calculations of grayscale and gradient factors. This method fails to effectively solve the problems of texture interference and morphological differentiation in composite board detection. For example, composite board surfaces often exhibit large-area, continuously distributed rolling textures or wire drawing marks. These textures are also high-frequency signals. The aforementioned methods mainly focus on distinguishing random noise, lacking effective means to suppress this large-scale, continuously distributed background texture interference, which is easily misidentified as burrs. Summary of the Invention

[0005] To address the technical problems of the aforementioned related technologies, which struggle to suppress large-area background texture interference and lack analysis of defect morphology characteristics, thus making it impossible to accurately distinguish between real burrs and noise points, this invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for detecting edge burrs on a composite plate, comprising: extracting the longest edge of a composite plate image as an edge baseline; performing morphological opening and closing operations on the composite plate image using a structuring element larger than the estimated burr width, and subtracting the opening and closing operation results from the composite plate image to obtain a first enhanced image and a second enhanced image; performing binarization processing on the first enhanced image and the second enhanced image and performing connected component analysis to obtain candidate regions; determining near-edge regions based on the distance between the center point coordinates of the candidate regions and the edge baseline; obtaining the eigenvalues ​​of the covariance matrix of the pixel coordinates within the near-edge regions, and obtaining the anisotropy index of the near-edge regions based on the eigenvalues; Near-edge regions are filtered based on the anisotropy index, and a first refined image and a second refined image are obtained. Pixels on the edge baseline are used as target pixels, and their grayscale values ​​are accumulated on the first and second refined images along the normal vector and its opposite direction to obtain a first energy and a second energy. The first energy and the second energy are added together to obtain the total energy. The total energy of all pixels on the edge baseline constitutes a total energy sequence. Points with energy values ​​below the noise threshold in the total energy sequence are set to zero, and consecutive sequence segments are searched. The defect index of the sequence segment is obtained based on its length, the sum of the corresponding energies, and the maximum value in the sequence segment. Burr defects are determined based on the defect index.

[0007] This invention achieves specific enhancement of bright and dark burr structures in composite board images by performing morphological opening and closing operations using structuring elements larger than the estimated burr width and by utilizing positional subtraction operations. Furthermore, this invention leverages the significant difference in morphological scale between burrs and rolling textures or wire drawing marks on the board surface to effectively suppress continuous, large-area background texture interference. Subsequently, this invention obtains anisotropy index based on the eigenvalues ​​of the covariance matrix of pixel coordinates in the near-edge region to describe the region's extension characteristics. This accurately distinguishes between real burr defects that tend to extend in morphology and isolated noise that appears as near-circular noise points, further refining the burr signal. Finally, by projecting the refined two-dimensional burr information onto the edge baseline to form a one-dimensional total energy sequence, and using the defect index to analyze continuous energy sequence segments, this invention achieves a measurement from two-dimensional image features to one-dimensional defect severity. It can stably distinguish between real burr defects and individual energy spike noise, significantly improving the accuracy and reliability of composite board edge burr defect detection.

[0008] Preferably, the step of extracting the longest edge of the composite plate image as the edge baseline includes: performing adaptive histogram equalization processing on the composite plate image to obtain a contrast-enhanced image; using local adaptive thresholding to segment the contrast-enhanced image to obtain a binary image; performing a contour search algorithm on the binary image to identify all closed contours in the image; and selecting the longest outer contour from all identified contours as the edge baseline.

[0009] Preferably, the step of subtracting the opening and closing operation results from the composite plate image to obtain the first enhanced image and the second enhanced image includes: subtracting the opening operation result from the composite plate image to obtain the first enhanced image; and subtracting the closing operation result from the composite plate image to obtain the second enhanced image.

[0010] This invention extracts bright burrs smaller than the structuring element in the image by subtracting the opening operation result from the composite plate image; at the same time, it extracts dark burrs smaller than the structuring element in the image by subtracting the closing operation result from the composite plate image, ensuring that both bright and dark burr defects can be effectively highlighted.

[0011] Preferably, determining the near-edge region based on the distance between the center point coordinates of the candidate region and the edge baseline includes: obtaining the center point coordinates of each candidate region, and selecting the candidate region whose center point coordinates are less than three times the size of the structural element as the near-edge region.

[0012] Preferably, the step of filtering near-edge regions based on the anisotropy index and obtaining a first refined image and a second refined image includes: taking the near-edge regions in the first enhanced image where the anisotropy index is greater than a preset threshold as the first spur region; setting the pixels outside the first spur region in the first enhanced image to 0 to obtain the first refined image; taking the near-edge regions in the second enhanced image where the anisotropy index is greater than a preset threshold as the second spur region; setting the pixels outside the second spur region in the second enhanced image to 0 to obtain the second refined image.

[0013] This invention filters near-edge regions by setting an anisotropy index threshold, retaining regions with anisotropy indices greater than the threshold as spur regions, and setting the pixels of other regions to 0. It utilizes the characteristic that real spurs tend to extend in shape, while noise points mostly present a circular shape. Through this morphological feature description, the spur signal is further purified, effectively eliminating noise points that do not conform to the spur morphological characteristics, and generating a first refined image and a second refined image for subsequent energy analysis.

[0014] Preferably, the heterogeneity index satisfies the following relationship: In the formula, For the first Anisotropy index of near-edge regions, and The first The maximum and minimum eigenvalues ​​of the covariance matrix of the horizontal and vertical coordinates of pixels within a near-edge region.

[0015] This invention utilizes the maximum eigenvalue of the covariance matrix representing the maximum extension of a region and the minimum eigenvalue representing the minimum extension of a region to obtain anisotropy indices, which can accurately describe the morphological characteristics of burrs. For elongated burrs, the ratio of the minimum to the maximum eigenvalues ​​approaches 0; while for circular noise, this ratio approaches 1, achieving an effective evaluation of the extension characteristics of burrs in near-edge regions.

[0016] Preferably, the step of accumulating the grayscale values ​​of the pixel along the normal vector and the opposite direction of the target pixel on the first refined image and the second refined image to obtain the first energy and the second energy includes: extracting a pixel sequence of a preset length on the first refined image along the normal vector of the target pixel, and accumulating the grayscale values ​​of the pixels in the sequence to obtain the first energy; and extracting a pixel sequence of a preset length on the second refined image along the opposite direction of the normal vector of the target pixel, and accumulating the pixel values ​​of the sequence to obtain the second energy.

[0017] Preferably, the step of setting the points in the total energy sequence whose energy values ​​are lower than the noise threshold to zero and then searching for consecutive sequence segments includes: setting the points in the total energy sequence whose energy values ​​are lower than the noise threshold to zero, and searching for all consecutive sequence segments in the processed total energy sequence whose energy values ​​are higher than the noise threshold.

[0018] Preferably, the defect index satisfies the following relationship: In the formula, For the first Defect index of each sequence segment, For the first The sum of the energies within each sequence segment, For the first The length of each sequence segment For the first The maximum energy within a sequence segment.

[0019] This invention combines the average energy and maximum energy of a sequence segment to obtain a defect index. The average energy reflects the overall intensity of burr defect features at the corresponding edge of the sequence segment, while the maximum energy reflects whether there are sharp burr defects at that location. By combining the two, the defect index can more comprehensively assess the severity of burr defects, providing a reliable basis for the final defect determination.

[0020] Secondly, the present invention provides a composite board edge burr detection system, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned composite board edge burr detection method is implemented.

[0021] By adopting the above technical solution, a computer program for detecting edge burrs on composite panels is generated and stored in a memory for loading and execution by a processor. This allows for the creation of a terminal device based on the memory and processor, facilitating its use.

[0022] The beneficial effects of this invention are as follows: Firstly, by utilizing the significant difference in morphological scale between burrs and the background texture of the composite board surface, and employing morphological opening and closing operations using structuring elements larger than the estimated maximum burr width, combined with positional subtraction, this invention achieves specific extraction of bright and dark burr signals. While enhancing the burr signal, it suppresses background texture interference with continuous, large-area distribution characteristics. Secondly, addressing the problem of isolated noise points that may still exist in the image after morphological enhancement, this invention further introduces morphological feature analysis. By calculating the anisotropy index of the near-edge region, it can utilize the true burr... The difference in structural morphology between burr defects and near-circular noise allows for secondary purification of the burr signal, effectively eliminating noise interference that does not conform to the burr extension morphology. Finally, this invention converts the two-dimensional refined burr information, after scale and morphology screening, into a one-dimensional total energy sequence by projecting it along the normal direction of the edge baseline. By analyzing the continuous energy response segments in the total energy sequence and combining it with the defect index for evaluation, this invention can separate real burr defects from residual isolated noise, ultimately achieving high-precision and high-reliability detection of edge burrs on composite boards under complex background textures and random noise interference. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a method for detecting edge burrs on composite panels according to the present invention;

[0024] Figure 2 This is a schematic illustration of the composite plate in the present invention;

[0025] Figure 3 This is a schematic illustration of the first enhanced view in the present invention;

[0026] Figure 4 This is a schematic illustration of the first refining diagram in the present invention;

[0027] Figure 5 This is a schematic illustration of the detection results in this invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0030] This invention discloses a method for detecting edge burrs on composite panels, referring to... Figure 1 This includes steps S1-S5:

[0031] S1. Image acquisition and edge baseline positioning.

[0032] It should be noted that in order to analyze the burrs on the edge of the composite board, it is necessary to eliminate the interference of the complex surface background of the composite board. Therefore, this invention extracts the stable and continuous contour edge in the composite board image. This contour will be used as the baseline for all subsequent burr analysis, thereby ignoring the fine texture of the composite board surface.

[0033] Specifically, a high-resolution area scan camera is used to acquire images of the composite plate. Adaptive histogram equalization is performed on the composite plate images to enhance local image contrast and suppress fine surface textures and random noise, resulting in a contrast-enhanced image. Local adaptive thresholding is then applied to the contrast-enhanced image to binarize it into foreground and background, yielding a binary image. A contour-finding algorithm is then performed on the binary image to identify all closed contours. From all identified contours, the longest outer contour is selected as the edge baseline.

[0034] S2. Perform burr morphological enhancement and texture suppression on the composite plate image to obtain the first enhancement image and the second enhancement image.

[0035] It should be noted that, visually, burrs appear as relatively small, light or dark raised structures attached to the surface of the board. In contrast, while rolling textures or brush marks on the board surface are also high-frequency signals, their morphology is usually continuous and distributed over a large area. Therefore, this invention utilizes the difference in morphological scale between burrs and background textures to achieve specific enhancement of the burr signal, making the detection of burrs on the edges of composite boards more accurate.

[0036] Specifically, a structuring element larger than the estimated maximum burr width is set. A morphological opening operation is performed on the composite plate image using this structuring element to obtain a first background image. The composite plate image is then subtracted from the first background image to obtain a first enhanced image. A morphological closing operation is performed on the composite plate image using the structuring element to obtain a second background image. The second background image is then subtracted from the composite plate image to obtain a second enhanced image.

[0037] It should be added that morphological opening is a process of erosion followed by dilation. Its physical meaning is to remove all bright structures smaller than the structuring element in the image, while retaining the main body of the board, reflective gradients, and large-scale background such as rolling texture. Similarly, morphological closing can remove all dark structures smaller than the structuring element in the image. Then, during the subtraction process, the content in the background estimation image is removed from the composite board image, achieving targeted enhancement of bright and dark burr structures.

[0038] S3. Obtain each candidate region based on the spur enhancement image, determine the near-edge region based on the distance between the center point coordinates of the candidate region and the edge baseline, obtain the anisotropy index of the near-edge region based on the coordinates of the pixels in the near-edge region, and construct the first refined image and the second refined image.

[0039] It should be noted that while the first and second enhancement images can effectively highlight burrs, in practical applications, some isolated noise points may still be retained, such as dust or small, strong textures. Real burr defects are not only convex in shape, but their structure also typically tends to extend in a specific direction rather than appearing as rounded noise. Therefore, this invention further purifies the burr signal through connected component analysis and morphological feature description, retaining only candidate regions that conform to the characteristics of burr defects in both shape and spatial location.

[0040] Specifically, the first enhanced image is first binarized to obtain a binary image. Next, connected component analysis is performed on the binary image to obtain candidate regions. The center point coordinates of each candidate region are obtained. Candidate regions whose center point coordinates are less than a distance threshold from the edge baseline are considered near-edge regions. The covariance matrix between the x and y coordinates of pixels within each near-edge region is obtained, and two eigenvalues ​​of the covariance matrix are acquired. The anisotropy index of the near-edge region is obtained based on these two eigenvalues. Similarly, the anisotropy index of each near-edge region in the second enhanced image is obtained.

[0041] For example, the distance threshold is 3 times the size of the structuring element.

[0042] Specifically, the heterosex index satisfies the following relation:

[0043] ;

[0044] In the formula, For the first Anisotropy index of near-edge regions, and The first The maximum and minimum eigenvalues ​​of the covariance matrix of the horizontal and vertical coordinates of pixels within a near-edge region.

[0045] in, Representing the The maximum spatial extension of a pixel within a near-edge region Representing the The minimum spatial extension of a pixel within a near-edge region; for a region corresponding to a thin, elongated burr, its maximum extension should be much greater than its minimum extension; while for circular noise, the extension in all directions is similar, therefore... As an index of opposite sexes The smaller the hour, the more... The closer a region is to the edge, the more likely it is to correspond to a fine, elongated burr. The greater the anisotropy index of a near-edge region, the higher the anisotropy index. When it is larger, the first The closer a region is to the edge, the more likely it is to be a circular noise region. The smaller the heterogeneity index of a region near its edge.

[0046] Furthermore, the near-edge region in the first enhanced image where the anisotropy index is greater than the threshold is taken as the first spur region, and all pixels outside the first spur region in the first enhanced image are set to 0 to obtain the first refined image; the near-edge region in the second enhanced image where the anisotropy index is greater than the threshold is taken as the second spur region, and all pixels outside the second spur region in the second enhanced image are set to 0 to obtain the second refined image, wherein the threshold is 0.6, and the implementer can adjust the threshold according to the actual situation.

[0047] For example, Figure 2 This is an image of the composite plate in this invention. Figure 3 This is the first enhanced diagram in the present invention. Figure 4 The first refined image in this invention shows that a large number of texture details unrelated to burr detection in the composite plate image have been removed from the first enhanced image. The first refined image further removes texture details unrelated to burr detection in the composite plate image based on the first enhanced image, making the burr defects more prominent in the image and refining the burr features of the image.

[0048] S4. Obtain the spur energy and total energy sequence based on the pixel sequence in the normal direction of each pixel point on the edge baseline and the pixel sequence in the opposite direction of the normal direction.

[0049] It should be noted that refined burr information distributed throughout the image can be obtained through the first and second refined images. However, the degree of burr defects at the edge of the composite plate cannot be measured directly using only the first and second refined images. Therefore, this invention obtains the burr energy based on the pixel sequence along the normal direction of each pixel on the edge baseline and the pixel sequence along the opposite direction of the normal direction.

[0050] Specifically, taking any pixel on the edge baseline as the target pixel, calculate the unit normal vector and the reverse normal vector of the target pixel. Along the unit normal vector, extract a pixel sequence of a preset length from the first refined image. By accumulating the grayscale values ​​of the pixels in this sequence, obtain the first energy of the target pixel. Similarly, along the reverse normal vector, extract a pixel sequence of a preset length from the second refined image. By accumulating the pixel values ​​in this sequence, obtain the second energy of the target pixel. Add the first energy and the second energy of the target pixel to obtain the total energy of the target pixel. Finally, arrange the total energies of all pixels on the edge baseline in their order on the edge baseline to obtain the total energy sequence.

[0051] For example, the preset length is 10.

[0052] It should be added that the first refined image is obtained based on morphological opening operations and positional subtraction, which mainly enhances the bright-colored structural burrs on the edge of the composite board; the second refined image is obtained based on morphological closing operations and positional subtraction, which mainly enhances the dark-colored burr structure. Since bright-colored burrs usually appear as protrusions towards the outside of the board, while dark-colored defects often appear inward or in the opposite shape, this invention achieves directional accumulation of energy for different types of burr defects by pairing the normal vector with the first refined image and pairing the normal vector in the opposite direction with the second refined image, ensuring that the total energy can comprehensively reflect the defect condition at the edge baseline.

[0053] S5. Merge the continuous energy responses in the total energy sequence into sequence segments and obtain the defect index to determine whether the burr segment constitutes a real burr defect.

[0054] It should be noted that the total energy sequence maps the spur information of a two-dimensional image to a one-dimensional space. In this sequence, real spur defects typically appear as continuous energy segments occupying a certain width in the total energy sequence, while isolated noise may appear as a single energy spike. Therefore, this invention merges the continuous energy responses in the total energy sequence into sequence segments and obtains a defect index to determine whether a spur segment constitutes a real spur defect.

[0055] Specifically, points in the total energy sequence with energy values ​​below the noise threshold are set to zero to filter out baseline noise. All consecutive sequence segments with energy values ​​above the noise threshold are then searched in the processed total energy sequence. The defect index of a sequence segment is obtained based on the length of each segment, the sum of the corresponding energies of the segment, and the maximum value in the segment.

[0056] For example, the noise threshold is 10.

[0057] Specifically, the defect index satisfies the following relationship:

[0058] ;

[0059] In the formula, For the first Defect index of each sequence segment, For the first The sum of the energies within each sequence segment, For the first The length of each sequence segment For the first The maximum energy within a sequence segment.

[0060] in, Representing the The average energy of the nth sequence segment; the larger this value, the better the nth sequence segment. The more likely there are strong burr defect characteristics at the edge baseline corresponding to the first sequence segment, the more likely there are to be the second sequence segment. The larger the defect index of the first sequence segment, the smaller the value indicates that the first... The more likely the edge baseline corresponding to the first sequence segment is to be free of burr defects, the more likely the second sequence segment will be free of defects. The smaller the defect index of a sequence segment.

[0061] Representing the The maximum energy of the nth sequence segment; the larger this value, the better the nth sequence segment. The more likely there are sharp burr defects at the edge baseline corresponding to the first sequence segment, the more likely there are to be such defects. The larger the defect index of the first sequence segment, the smaller the value indicates that the first... The more likely there is a gentle composite plate edge at the edge baseline corresponding to the first sequence segment, the more likely there is to be a second one. The smaller the defect index of each sequence segment.

[0062] Furthermore, the edge baseline corresponding to the sequence segment with a defect index greater than the defect threshold is identified as a burr defect. The implementer can determine whether the composite board meets the process standard based on the number of burr defects. For example, if the number of burr defects is greater than 10, the composite board does not meet the process standard.

[0063] For example, the defect threshold is 50.

[0064] For example, Figure 5 The results of the present invention show that the present invention can eliminate the interference of the surface texture of the composite board and can better detect the burrs on the edge of the composite board.

[0065] This invention also discloses a composite board edge burr detection system, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a composite board edge burr detection method according to the present invention.

[0066] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

Claims

1. A method of detecting edge burrs of a composite panel, characterized by, The method comprises the following steps: The longest edge of the composite board image is extracted as an edge reference line; a morphological opening and closing operation is performed on the composite board image using a structure element larger than the estimated burr width, and the composite board image is subtracted from the opening and closing operation results respectively to obtain a first enhanced image and a second enhanced image; The first enhanced image and the second enhanced image are subjected to binaryzation processing and connected domain analysis to obtain candidate regions; near-edge regions are determined according to the distance between the center point coordinates of the candidate regions and the edge reference line; The eigenvalues of the covariance matrix of the pixel point coordinates in the near-edge regions are obtained, and the anisotropy index of the near-edge regions is obtained according to the eigenvalues, which satisfies the relationship: ； In the formula, is the first is the second is the third is the fourth is the fifth The near-edge regions are screened according to the anisotropy index, and a first refined image and a second refined image are obtained. The pixel points on the edge reference line are taken as target pixel points, the gray values of the pixel points on the first refined image and the second refined image are accumulated along the normal vector and the opposite direction of the normal vector of the target pixel points respectively to obtain a first energy and a second energy; the first energy and the second energy are added to obtain a total energy; the total energy of all pixel points on the edge reference line forms a total energy sequence; after the points with energy values lower than a noise threshold in the total energy sequence are set to zero, a continuous sequence segment is searched; The defect index of the sequence segment is obtained according to the length of the sequence segment, the sum of the energy corresponding to the sequence segment and the maximum value in the sequence segment, which satisfies the relationship: ； wherein is the defect index of the th sequence segment, is the sum of the energies in the th sequence segment, is the length of the th sequence segment, is the maximum value of the energies in the th sequence segment; The burr defect is determined according to the defect index.

2. The method of claim 1, wherein, The longest edge of the composite board image is extracted as an edge reference line, which comprises the following steps: adaptive histogram equalization processing is performed on the composite board image to obtain a contrast-enhanced image; a local adaptive threshold segmentation is performed on the contrast-enhanced image to obtain a binary image, and a contour finding algorithm is performed on the binary image to identify all closed contours in the image; from all the identified contours, the longest external contour is selected as the edge reference line.

3. The method of claim 1, wherein, The composite board image is subtracted from the opening and closing operation results respectively to obtain a first enhanced image and a second enhanced image, which comprises the following steps: the composite board image is subtracted from the opening operation result to obtain the first enhanced image; the composite board image is subtracted from the closing operation result to obtain the second enhanced image.

4. The method of claim 1, wherein, The near-edge regions are determined according to the distance between the center point coordinates of the candidate regions and the edge reference line, which comprises the following steps: the center point coordinates of each candidate region are obtained; the candidate regions with a distance between the center point coordinates and the edge reference line less than three times the size of the structure element are taken as the near-edge regions.

5. The method of claim 1, wherein, The near-edge regions are screened according to the anisotropy index, and a first refined image and a second refined image are obtained, which comprises the following steps: the near-edge regions with an anisotropy index greater than a preset threshold in the first enhanced image are taken as first burr regions; the pixel points outside the first burr regions in the first enhanced image are set to 0 to obtain the first refined image; the near-edge regions with an anisotropy index greater than a preset threshold in the second enhanced image are taken as second burr regions, and the pixel points outside the second burr regions in the second enhanced image are set to 0 to obtain the second refined image.

6. The method of claim 1, wherein, The first energy and the second energy are obtained by accumulating the gray values of the pixels on the first refined image and the second refined image along the normal vector of the target pixel and the reverse direction of the normal vector, and the method comprises the following steps: extracting a pixel sequence with a preset length on the first refined image along the normal vector of the target pixel, accumulating the gray values of the pixels in the sequence, and obtaining the first energy; extracting a pixel sequence with a preset length on the second refined image along the reverse direction of the normal vector of the target pixel, accumulating the pixel values of the sequence, and obtaining the second energy.

7. The method of claim 1, wherein, After the points with energy values lower than the noise threshold in the total energy sequence are set to zero, the continuous sequence segments are searched, and the method comprises the following steps: setting the points with energy values lower than the noise threshold in the total energy sequence to zero, and searching all continuous sequence segments with energy greater than the noise threshold on the processed total energy sequence.

8. A composite panel edge burr detection system, characterized by, The method comprises the following steps: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a composite board edge burr detection method according to any one of claims 1-7 is realized.

Images