A method, apparatus, and medium for manufacturing a dot-sol gel fabric

Abstract

The application provides a point-sol gel fabric manufacturing method, equipment and medium, relates to the point-sol gel fabric manufacturing technical field, and comprises the following steps: dividing a target image into a plurality of sub-images; performing DCT transformation on each sub-image to obtain a corresponding DCT coefficient matrix of each sub-image; clustering all DCT coefficient matrices according to the characteristics of each DCT coefficient matrix; processing each DCT coefficient matrix in QA to obtain a processed DCT coefficient matrix cluster WA corresponding to QA; mapping the energy center corresponding to each DCT coefficient matrix in WA to the target image to obtain the centroid of each DCT coefficient matrix in QA; obtaining the centroid of a complete glue point composed of a defective glue point; determining whether the point-sol gel fabric to be detected is abnormal according to the centroid distribution of each glue point in the target image; and obtaining a final fabric; and the application can greatly reduce the data processing amount and significantly improve the detection efficiency.

Description

Technical Field

[0001] This invention relates to the field of dot-sol fabric manufacturing technology, and in particular to a method, equipment and medium for manufacturing dot-sol fabric. Background Technology

[0002] Dot-bonded fabric, as a functional composite material, is widely used in medical protective clothing, outdoor apparel, and other fields by uniformly distributing adhesive dots on the surface of a base fabric and bonding them under heat and pressure. In the manufacturing of dot-bonded fabric, to ensure the integrity, uniformity, and accuracy of the adhesive dots, the fabric typically needs to be inspected. Existing inspection methods usually involve inspecting the entire image of the fabric, i.e., identifying adhesive dots through template matching or threshold segmentation, and then inspecting each dot individually. This method requires traversing all pixels, resulting in slow processing speeds and making it difficult to meet the real-time inspection requirements of high-speed production lines. Summary of the Invention

[0003] To address the aforementioned technical problems, the technical solution adopted by this invention is as follows:

[0004] According to a first aspect of this application, a method for manufacturing a dot-sol fabric is provided, the method comprising the following steps:

[0005] S100, acquire the target image corresponding to the sol-gel fabric point to be detected; the target image contains several glue points;

[0006] S200, based on the number of pixels corresponding to each glue dot in the target image and the resolution of the target image, divide the target image into several sub-images; among these sub-images, there are sub-images that contain only one complete glue dot;

[0007] S300, perform DCT transformation on each sub-image to obtain the DCT coefficient matrix corresponding to each sub-image;

[0008] S400, based on the characteristics of each DCT coefficient matrix, cluster all DCT coefficient matrices to obtain a cluster QA of DCT coefficient matrices containing a complete glue dot;

[0009] S500, process each DCT coefficient matrix in QA to obtain the processed DCT coefficient matrix cluster WA corresponding to QA; wherein, each DCT coefficient matrix in WA contains only the DCT coefficients corresponding to the glue dot image;

[0010] S600 maps the energy center corresponding to each DCT coefficient matrix in WA to the target image to obtain the centroid of the glue point corresponding to each DCT coefficient matrix in QA.

[0011] S700, stitch together the sub-images corresponding to each DCT coefficient matrix that is not in QA in the target image to obtain the centroid of the complete glue dot composed of incomplete glue dots;

[0012] S800 determines whether there are any abnormalities in the sol-gel fabric at the point to be detected based on the centroid distribution of each glue dot in the target image.

[0013] S900: If there are no abnormalities in the sol-bonded fabric at the test point, hot pressing is performed to obtain the final fabric.

[0014] According to another aspect of this application, a non-transitory computer-readable storage medium is also provided, wherein at least one instruction or at least one program is stored in the storage medium, and the at least one instruction or at least one program is loaded and executed by a processor to implement the above-described method for manufacturing dot-sol fabric.

[0015] According to another aspect of this application, an electronic device is also provided, including a processor and the aforementioned non-transitory computer-readable storage medium.

[0016] The present invention has at least the following beneficial effects:

[0017] The manufacturing method of dot-bonded fabric of the present invention divides the target image into sub-images containing complete glue dots, incomplete glue dots, or no glue dots according to the number of glue dots pixels and the image resolution. This avoids the redundant calculation of processing a massive number of pixels required by existing full-image traversal detection, focuses the detection range on the glue dot-related area, greatly reduces the amount of data processing, and significantly improves the detection efficiency, which can meet the stringent requirements of high-speed production lines for real-time detection.

[0018] Furthermore, by using DCT coefficient matrix feature clustering, and leveraging the differences in energy distribution in the frequency domain for different glue dot states, coefficient matrix clusters containing complete glue dots, incomplete glue dots, and glue-free dots are accurately separated. This effectively solves the problems of complete and incomplete glue dots and background noise aliasing in traditional methods, improving the purity and accuracy of glue dot feature extraction. Through energy center mapping and incomplete glue dot splicing, only the DCT coefficients corresponding to the glue dots are retained and mapped to the spatial domain to determine the centroid. At the same time, the segmented incomplete glue dot sub-images are spliced ​​and reconstructed, eliminating the interference of background noise on centroid localization and compensating for the localization deviation caused by glue dot incompleteness or segmentation, achieving high-precision localization of glue dot centroids. Based on multi-dimensional analysis of centroid distribution, combined with topological features and spatial pattern recognition, subtle changes in glue dot integrity, distribution uniformity, and positional accuracy can be sensitively captured, effectively identifying local anomalies and minor defects that are difficult to detect by traditional methods, significantly improving the sensitivity and reliability of detection. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart illustrating a method for manufacturing a dot-sol fabric according to an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] It should be noted that, based on this disclosure, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects set forth herein can be used to implement the device and / or practice the method. Furthermore, this device and / or practice the method can be implemented using other structures and / or functionalities besides one or more of the aspects set forth herein.

[0023] The following will refer to Figure 1 The flowchart shown illustrates a method for manufacturing a dot-sol fabric, introducing such a method.

[0024] The method for manufacturing this dot-sol fabric may include the following steps:

[0025] S100, acquire the target image corresponding to the sol-gel fabric at the point to be detected; the target image contains several glue dots.

[0026] In this embodiment, after applying adhesive to the fabric using non-contact jet dispensing technology, a sol-gel fabric to be tested is obtained. The sol-gel fabric to be tested has several adhesive dots arranged in a matrix. An industrial camera can be used to photograph the sol-gel fabric to be tested to obtain a target image corresponding to the sol-gel fabric to be tested. The target image can be a grayscale image. Grayscale conversion of the image can significantly reduce the computational load of subsequent DCT transformation and feature extraction.

[0027] S200: Based on the number of pixels corresponding to each glue dot in the target image and the resolution of the target image, the target image is divided into several sub-images; among these sub-images, there exists a sub-image containing only one complete glue dot.

[0028] In this embodiment, based on the number of glue dots and image resolution, the target image is divided into several sub-images. Each sub-image may contain complete glue dots, incomplete glue dots, or no glue dots. This step can reduce the computational unit and decrease the complexity of subsequent DCT transformation and feature analysis. It can also enhance local feature extraction, perform refined analysis on individual glue dots or local areas, and reduce background interference. Furthermore, it can adapt to glue dots of different sizes by dynamically adjusting the sub-image resolution according to the glue dot size to ensure complete capture of the largest glue dot.

[0029] Furthermore, step S200 may include the following steps:

[0030] S210, obtain the longest side of the smallest bounding box corresponding to each glue dot in the target image, to obtain a list of longest sides H = (H1, H2, ..., H...). i H n ), i = 1, 2, ..., n; where H i Let n be the longest side of the rectangular bounding box corresponding to the i-th glue dot, and n be the number of glue dots.

[0031] For each glue dot in the target image, calculate its minimum bounding rectangle, which is the smallest rectangle that exactly encloses the outline of the glue dot, and extract the longest side of the rectangle to obtain H.

[0032] The longest side of the minimum bounding box objectively reflects the geometric size of the glue dot, providing a data basis for the unified setting of the sub-image resolution and avoiding unreasonable sub-image division due to differences in glue dot size. The irregular shape of the glue dot is transformed into a rectangular feature, which facilitates subsequent batch processing and improves the universality and robustness of the algorithm.

[0033] S220, get the target longest side MH=MAX(H); MAX() is the preset maximum value function.

[0034] Using the size of the largest glue dot as a benchmark, we ensure that the resolution of subsequent sub-images is sufficient to accommodate all glue dots, avoiding glue dot information loss, such as incompleteness or truncation, due to the factor image being too small. In addition, using the global maximum size as a reference facilitates the unified planning of the size of subsequent sub-images and reduces the processing complexity caused by size differences.

[0035] S230, obtain the number of pixels NUM1 corresponding to MH.

[0036] Since MH is based on the length of image pixels, such as the number of pixels, we directly count the number of pixels corresponding to MH, NUM1, which is used for subsequent resolution quantization calculations.

[0037] Converting geometric dimensions into image pixel units ensures that the resolution setting matches the actual image data precision, guaranteeing the accuracy of sub-image segmentation. NUM1 serves as the numerical basis, providing a quantitative foundation for the power-law processing of sub-image resolution, ensuring the scientific nature of the size setting.

[0038] S240, if NUM1=2 r Therefore, the resolution of the sub-image is determined to be 2. r ×2 r ; r is an integer greater than 0.

[0039] If NUM1 is exactly a power of 2, such as NUM1 = 8 = 2³, and r = 3, then the resolution of the sub-image is set to 2. r ×2 r For example, 8×8 pixels, ensure that the resolution is a square and the side length is a power of 2.

[0040] Discrete Cosine Transform (DCT) typically requires the input block size to be a power of 2, such as 8×8 or 16×16. This setting directly accommodates the computational requirements of DCT, avoiding additional size adjustments or padding operations and improving computational efficiency.

[0041] To ensure information integrity: A square sub-image with a side length of NUM1 can fully accommodate the largest glue dot, while ensuring that other smaller glue dots have boundary space in the sub-image to avoid the glue dot edge information being truncated.

[0042] S250, if 2 r <NUM1<2 r+1 Therefore, the resolution of the sub-image is determined to be 2. r+1 ×2 r+1 .

[0043] If NUM1 lies between two consecutive powers of 2, such as NUM1=10, then it lies between 2³=8 and... If the value is between 2 and 2, then round up to the next power of 2, setting the sub-image resolution to 2. r+1 ×2 r+1 For example, 16×16 pixels.

[0044] To ensure that even if the glue dot size is not a power of 2, the sub-image can still completely contain the glue dot by rounding up, avoiding the loss of glue dot information due to insufficient size and improving the robustness of the algorithm; a unified computational framework is used, and the sub-image resolution is always a power of 2 regardless of the glue dot size, which facilitates the batch processing of subsequent DCT transformations, such as hardware acceleration or parallel computing, and reduces the complexity of the algorithm implementation; a balance is struck between computational efficiency and accuracy, avoiding glue dot defects due to excessive pursuit of small sizes, or computational resource waste due to excessively large sizes, achieving the optimal balance between information integrity and computational efficiency in resolution setting.

[0045] S300, perform DCT transformation on each sub-image to obtain the DCT coefficient matrix corresponding to each sub-image.

[0046] The discrete cosine transform (DCT) is performed on each sub-image to convert the spatial domain image into a frequency domain coefficient matrix. The DCT transform highlights the frequency characteristics of the image, concentrating energy in the low-frequency part and reflecting details in the high-frequency part.

[0047] DCT transform concentrates image energy on low-frequency coefficients, suppresses noise interference, and facilitates the extraction of essential features of glue dots, such as shape and texture. Compared with the original pixel data, the DCT coefficient matrix is ​​more compact, reducing the amount of subsequent calculations.

[0048] It should be noted that those skilled in the art can use existing DCT transformation methods to perform DCT transformation on each sub-image according to actual needs, so as to obtain the DCT coefficient matrix corresponding to each sub-image, which will not be elaborated here.

[0049] S400, based on the characteristics of each DCT coefficient matrix, cluster all DCT coefficient matrices to obtain a cluster QA of DCT coefficient matrices containing a complete glue dot.

[0050] In this embodiment, feature extraction can be performed on each DCT coefficient matrix. Feature extraction includes: low-frequency energy feature extraction and non-zero coefficient distribution feature extraction.

[0051] Low-frequency energy feature extraction: After DCT transformation, the upper left corner of the matrix represents low-frequency coefficients, indicating the overall image structure, while the lower right corner represents high-frequency coefficients, indicating details and edges. For sub-images containing glue dots, the proportion of low-frequency energy is higher for complete glue dots, while the proportion of low-frequency energy is lower for sub-images with incomplete glue dots or no glue dots.

[0052] Calculate the energy proportion of the low-frequency region for each DCT coefficient matrix: for example, take the first k×k low-frequency coefficients, such as the first 4×4 in an 8×8 matrix, and calculate the proportion of their energy to the total energy of the entire matrix; where the energy is the sum of squares of the coefficients.

[0053] Non-zero coefficient distribution characteristics: Sub-images with complete glue dots contain complete structures, resulting in a more concentrated and numerous distribution of non-zero values ​​in their DCT coefficient matrix; sub-images without glue dots have coefficients that are mostly zero or close to zero. Statistically analyze the number and spatial distribution density of non-zero coefficients in each matrix, such as the proportion of non-zero coefficients in high-frequency regions.

[0054] After extracting the above features, feature vectors are constructed. The features, such as the proportion of low-frequency energy, the number of non-zero coefficients, the proportion of high-frequency energy, and the mean / variance of the coefficient matrix, are combined into a multi-dimensional feature vector, which is used as the input for clustering.

[0055] Clustering algorithms can use K-means clustering or hierarchical clustering, which merges similar clusters from bottom to top, making it easier to cluster based on feature distance.

[0056] After clustering, the feature vectors of QA satisfy the following conditions: the low-frequency energy ratio is the highest, the number of non-zero coefficients is the largest, and the distribution is concentrated. Other feature vectors that are clustered into QA are not feature vectors corresponding to sub-images containing a complete glue dot. For example, DCT coefficient matrices containing at least one incomplete glue dot and DCT coefficient matrices that do not contain glue dots.

[0057] Furthermore, clustering results can be verified by manually labeling a small number of samples. If misclassification occurs, the feature weights or initial values ​​of cluster centers can be adjusted.

[0058] S500, process each DCT coefficient matrix in QA to obtain the processed DCT coefficient matrix cluster WA corresponding to QA; wherein, each DCT coefficient matrix in WA contains only the DCT coefficients corresponding to the glue dot image.

[0059] Background noise is removed from the DCT coefficient matrix containing complete glue dots, retaining only the feature information of the glue dots themselves, thus providing pure feature data for subsequent centroid calculation.

[0060] Furthermore, step S500 may include the following steps:

[0061] S510, Obtain the standard DCT coefficient matrix D of the non-sol-gel fabric corresponding to QA. bg ; where D bg The image was obtained by performing DCT transformation on the image of the non-sol fabric. The surface tension, material, and knitting method of the fabric were the same during the process of photographing the non-sol fabric and the fabric to be tested for the sol.

[0062] In this embodiment, the fabric without adhesive dots, i.e., the same type of fabric that has not undergone dot-sol treatment, is photographed, and its image is subjected to DCT transformation to obtain the standard DCT coefficient matrix D. bg It should be noted that the surface tension of fabric can change, and this change affects the fabric's texture. Similarly, the fabric's material and knitting method also influence its texture. Therefore, the surface tension, material, and knitting method of the fabric being photographed must be completely consistent with the fabric being tested to ensure D... bg It can accurately reflect the background characteristics of the fabric to be tested, namely the inherent texture and structure when there are no glue dots.

[0063] This step enables the establishment of a reliable background benchmark because D bg This represents the frequency domain characteristics of the fabric without adhesive dots, providing a reference template for subsequent background noise removal. If shooting conditions are inconsistent, D...bg It may contain variables that are unrelated to the background, such as fabric deformation and material differences, which may lead to errors in subsequent difference processing.

[0064] It can also improve feature separation accuracy by fixing shooting conditions to ensure D bg The high degree of matching with the background features of the image to be detected enables subsequent interpolation operations to accurately separate the feature signals of the glue dots.

[0065] S520, for any DCT coefficient matrix D in QA fg D fg With D bg Perform subtraction on the same digit to obtain D. fg With D bg The corresponding DCT coefficient difference matrix D diff =(DE1, DE2, ..., DE p , ..., DE q ), p=1,2,…,q; where DE p D fg With D bg The difference corresponding to the p-th coefficient, q is D fg With D bg The number of coefficients in the middle.

[0066] In this embodiment, the DCT coefficient matrix D for each complete glue dot in QA is... fg With standard DCT coefficient matrix D bg Subtracting the corresponding position coefficients yields the difference matrix D. diff D fg It includes background features and glue dot features, while D bg Only background features are included, therefore, D diff The essence is the glue dot characteristic, and the difference mainly reflects the contribution of the glue dots to the frequency domain coefficients.

[0067] Through the difference operation, background features, such as the frequency domain coefficients of the inherent texture of the fabric, are canceled out, and the difference matrix D... diff It focuses more on the frequency domain components corresponding to the shape, edge and other features of the glue dots, providing a foundation for subsequent noise reduction; it reduces background interference and avoids the mixing of background texture and glue dot features, so that subsequent processing can directly target the essential features of the glue dots and improve the accuracy of subsequent centroid calculation.

[0068] S530, Traverse D diff For each coefficient in the equation, if -η < DE p If <η, then D fg Set the p-th coefficient in the matrix to 0 to obtain D. fg The corresponding processed DCT coefficient matrix; η is the preset DCT coefficient difference threshold.

[0069] Set a threshold η, and traverse the difference matrix D. diff Each coefficient DE p . If DE p If the absolute value is less than η, the coefficient difference is considered to be caused by background noise or small fluctuations, rather than by glue dot characteristics; D fg The coefficients at the corresponding positions are set to 0 to eliminate the influence of background noise, and the coefficients that are finally retained correspond to the significant features of the glue dots.

[0070] η can be determined by the following method:

[0071] Step 1: Collect standard data for non-sol-gel fabric:

[0072] Multiple sets of images of non-sol-bonded fabrics (with the same material, tension, and knitting method as the fabric to be tested) were taken, and DCT transformation was performed on each set of images to obtain multiple standard DCT coefficient matrices.

[0073] Step 2: Calculate the statistical distribution of the background coefficient:

[0074] The co-position coefficients of all standard DCT coefficient matrices are statistically analyzed, and the mean μ and standard deviation σ of each co-position coefficient are calculated.

[0075] Step 3: Determine the empirical threshold for η:

[0076] η can be set to k×σ, where k is an empirical coefficient ranging from 2 to 3, to cover 95% to 99.7% of the background noise fluctuation range. For example, when k=2, η=2σ, which can filter out approximately 95% of the background noise coefficient.

[0077] The DCT matrix processed through the above steps contains only glue dot features, ensuring that subsequent operations such as centroid calculation and anomaly detection are based on pure data, thereby improving the quality stability and detection reliability of dot-bonded fabric manufacturing.

[0078] S600 maps the energy center corresponding to each DCT coefficient matrix in WA to the target image to obtain the centroid of the glue point corresponding to each DCT coefficient matrix in QA.

[0079] Furthermore, step S600 may include the following steps:

[0080] S610, for any DCT coefficient matrix D in WA e , obtain D e Corresponding total energy Where N-1 is D e The number of rows or columns; (u, v) is D e The frequency coordinates; E(u, v) is the energy of the coefficient corresponding to (u, v); E(u, v) = |D e (u, v) |2 ;D e (u, v) are the coefficients corresponding to (u, v).

[0081] In this embodiment, for any DCT coefficient matrix D in the sub-image set WA e Calculate its total energy E total In signal processing, energy is positively correlated with the square of the signal amplitude, and the sum of squares of the DCT coefficients reflects the total energy distribution of the information in the sub-image. The presence of glue dots will cause the coefficient energy of the corresponding region to be significantly higher than that of the background; therefore, the total energy can be used to characterize the energy intensity of the glue dots.

[0082] The total energy can measure the energy concentration of the glue dots in the frequency domain, providing a benchmark for subsequent positioning; the squaring operation amplifies the weight of high amplitude coefficients, weakens the influence of low energy noise, and improves the stability of subsequent calculations.

[0083] S620, based on E(u, v) and E total Determine D e Frequency coordinates of the energy center (u) center v center ); ; ;

[0084] In this embodiment, the energy center can be understood as the centroid of energy distribution in the frequency domain. The coefficient energy corresponding to the glue point is higher, which will cause the energy center to shift towards its frequency characteristics, thereby indirectly reflecting the position characteristics of the glue point in the spatial domain.

[0085] By using weighted averaging, low-energy interference in non-glued point regions is eliminated, making the center coordinates closer to the actual frequency characteristics of the glued points. In addition, the energy distribution is transformed into specific coordinates, providing clear mathematical parameters for subsequent spatial domain mapping.

[0086] S630, will (u center v center ) converted to D e The corresponding spatial coordinates within the sub-image are used to obtain D. e The centroid of the corresponding glue dot.

[0087] The energy center coordinates (u) in the frequency domain center v center The coordinates of the glue dots are converted into spatial coordinates within the sub-image, thus obtaining the centroid of the glue dots.

[0088] It should be noted that the coordinate mapping relationship of DCT transformation needs to be considered during the conversion. DCT transformation converts the N×N image in the spatial domain into a frequency domain matrix of the same size, where (u,v)=(0,0) corresponds to the DC component, i.e., the low-frequency background, and the high-frequency components are distributed at the edges of the matrix. The mapping between spatial coordinates and frequency coordinates needs to be combined with the sub-image resolution N, and through inverse transformation logic or geometric mapping relationship, such as central symmetry transformation, the spatial centroid coordinates of the glue point in the sub-image are finally obtained.

[0089] For example, assuming the sub-image size is N×N, the spatial domain position mapping of the frequency coordinates (u,v) can be derived through the basis function distribution of the inverse DCT transform, or directly using the linear correspondence between frequency and spatial position. Those skilled in the art can, according to actual needs, use existing transformation methods to map the energy center coordinates (u,v) in the frequency domain. center v center The coordinates are converted to spatial coordinates within the sub-image, which will not be elaborated here.

[0090] Through the above steps, the energy characteristics in the frequency domain are transformed into specific locations in the spatial domain, realizing the mapping from energy distribution to physical location, which directly serves the detection and adjustment of adhesive dots. The calculation of the energy center combines the weights of all coefficients, which is more resistant to local noise than directly taking the extreme point, making the centroid positioning more accurate and ensuring the uniformity of adhesive dot distribution and process precision of the sol-gel fabric. The precise centroid coordinates can be used to guide the subsequent adjustment of the sol-gel spraying equipment, reduce adhesive dot offset errors, and improve the consistency of fabric quality.

[0091] S700, stitch together the sub-images corresponding to each DCT coefficient matrix that is not in QA in the target image to obtain the centroid of the complete glue dot composed of incomplete glue dots.

[0092] In this embodiment, it is understood that since the size of the sub-image is based on the pixel data of the glue dot, each sub-image is likely to cover a complete glue dot. The number of sub-images containing incomplete glue dots and those not containing glue dots is in the minority. For such sub-images, traditional image processing methods can be used to extract the centroid of the glue dot.

[0093] Furthermore, step S700 may include the following steps:

[0094] S710: For each DCT coefficient matrix not in QA, the corresponding sub-image in the target image is stitched together to obtain a complete glue dot composed of several incomplete glue dots.

[0095] For each DCT coefficient matrix not included in the QA, it is first converted into a corresponding sub-image, either through inverse DCT transformation or by directly mapping it to a sub-region of the original image. Since the glue dots may be distributed in multiple adjacent sub-images, for example, if the glue dots are captured across regions, resulting in multiple incomplete parts, it is necessary to stitch together the parts belonging to the same glue dot in these sub-images to restore the complete shape of the glue dot.

[0096] The stitching process needs to be based on the position coordinates or feature matching of the glue dots. For example, the adjacent relationship can be determined by using the coordinate position of the sub-image in the original image, or the scattered fragments can be combined into a complete glue dot region by aligning the edge features of the glue dots.

[0097] By following the steps above, we can avoid the adhesive dots being truncated into multiple parts due to image segmentation, ensuring that subsequent analysis is based on the complete shape of the adhesive dots and reducing missed detections or false positives. For adhesive dots that span multiple regions, the splicing operation can restore their true shape, providing an accurate basis for subsequent contour extraction and centroid calculation, which is especially suitable for fabric detection with large or irregularly distributed adhesive dots.

[0098] S720, obtains the outline of each complete glue dot.

[0099] Contour extraction is performed on the stitched, complete glue dot image. Specific steps may include:

[0100] Binarization: Separate the glue dot area from the background to form a black and white image (the glue dot is the foreground and the background is the background).

[0101] Edge detection: Use algorithms such as Canny operator and Sobel operator to extract the boundary pixels of glue dots.

[0102] Contour tracing: Connects edge pixels sequentially to form a closed contour curve, used to describe the shape boundary of the glue dot.

[0103] Contours are the foundation of the geometric features of adhesive dots, such as area, perimeter, and roundness. Contours help determine whether adhesive dots conform to standard shapes, and whether there are any deformations or defects. Furthermore, they can distinguish adjacent adhesive dots. For densely distributed adhesive dots, contour extraction can clearly separate different dots, avoiding false merging due to adhesion and improving detection accuracy.

[0104] S730 determines the centroid of each complete glue dot based on its outline.

[0105] It should be noted that those skilled in the art can determine the centroid of each complete adhesive dot using existing centroid calculation methods according to actual needs, which will not be elaborated here.

[0106] The centroid can serve as a positioning reference for adhesive dots on the fabric, used to determine whether the distribution of adhesive dots is uniform and whether it deviates from the preset position, ensuring the accuracy of the dot-on-solution process. The degree of centroid offset can reflect the stability of adhesive dot coating, such as whether the positional deviation is caused by fabric tension or process error, providing a quantitative basis for quality control in the production process. The centroid coordinates can be used for calculating the spacing between adhesive dots, analyzing the arrangement pattern, etc., to help determine whether the fabric meets the design standards, such as the regularity and uniformity of the adhesive dot array.

[0107] S800 determines whether there are any abnormalities in the sol-gel fabric at the point to be detected based on the centroid distribution of each glue dot in the target image.

[0108] Furthermore, step S800 may include the following steps:

[0109] S810 converts the centroid of each glue dot into a two-dimensional spatial point set and constructs the corresponding Delaunay triangulation and Voronoi diagram for the two-dimensional spatial point set.

[0110] In this embodiment, each execution corresponds to coordinates, which can convert the centroid of each glue dot into a two-dimensional spatial point set; the Delaunay triangulation captures the spatial adjacency relationship between glue dots, and the Voronoi diagram quantifies the influence range of each glue dot, providing a mathematical basis for subsequent topological and geometric feature extraction; the triangulation and Voronoi diagram are based on the global topological structure, and the small deviation of the position of a single glue dot has little impact on the overall features, which is suitable for noisy environments in industrial scenarios.

[0111] S820, Obtain the topological features corresponding to the Delaunay triangulation; the topological features include: the number of adjacent nodes of each centroid, the average side length of the Delaunay triangulation, the standard deviation of the side length, and the distribution of the interior angles of the triangles.

[0112] In this embodiment, the number of adjacent nodes, i.e., the number of adjacent nodes of each centroid in the triangulation, reflects the local density. The average edge length, i.e., the average length of all edges in the triangulation, reflects the average distance between the glue points. The standard deviation of the edge length measures the dispersion of the edge length, reflecting the uniformity of the glue point distribution. The triangle interior angle distribution, i.e., the angular distribution of all triangle interior angles, assesses the regularity of the triangulation; ideally, all interior angles of equilateral triangles are 60°.

[0113] Topological features are used to determine whether the arrangement of adhesive dots meets design expectations, such as a uniform lattice. For example, an excessively large standard deviation may indicate uneven dot density; abnormal distribution at interior corners may indicate localized disorder. Topological features detect anomalies at the overall structural level, avoiding misjudgments at single points, and are suitable for identifying batch production defects.

[0114] S830, Based on the Voronoi diagram, extract the Voronoi polygon features corresponding to each centroid; the Voronoi polygon features include: the area, perimeter, and compactness of the polygon.

[0115] In this embodiment, the polygon area is obtained by calculating the area of ​​the Voronoi polygon corresponding to each centroid, reflecting the sphere of influence around the glue point; the polygon perimeter, i.e. the length of the quantized polygon boundary, indirectly reflects the adjacency complexity around the glue point; the compactness reflects the isotropic distribution of the glue point.

[0116] Area characteristics directly reflect the density around the adhesive dots; abnormally large or small areas may correspond to missing or overlapping adhesive dots. Compactness characteristics detect the regularity of the adhesive dot distribution; compactness deviating from normal values ​​may indicate local abnormalities, such as dot deformation caused by uneven surface tension.

[0117] S840 generates the glue dot feature vector XL corresponding to the target image based on the topological features corresponding to the Delaunay triangulation and the Voronoi polygon features corresponding to each centroid.

[0118] All features extracted in steps S820 and S830, such as the number of adjacent nodes, average edge length, and polygon area, are combined into a high-dimensional vector XL, with each feature corresponding to one dimension of the vector.

[0119] Complex topological and geometric features are compressed into a unified vector, which facilitates subsequent similarity calculation and anomaly detection. The feature vector provides a standardized description of the glue dot distribution, which can be directly compared with a preset standard, reducing the complexity of the algorithm.

[0120] S850, if the similarity between XL and the preset standard glue dot feature vector is greater than the preset similarity threshold, then it is determined that there is no abnormality in the sol-gel fabric at the point to be detected.

[0121] The feature vector XL is compared with the preset standard adhesive dot feature vector, and the similarity is calculated, such as cosine similarity or the reciprocal of Euclidean distance. If the similarity is greater than the preset threshold T, the adhesive dot distribution of the fabric is determined to be normal; otherwise, there is an anomaly, such as missing, overlapping, or unevenly arranged adhesive dots.

[0122] By comparing feature vectors, rapid and objective anomaly detection can be achieved, avoiding the subjectivity and inefficiency of manual detection. The similarity threshold T can be dynamically adjusted according to production process requirements to adapt to different accuracy needs, such as higher thresholds required for high-end fabrics.

[0123] Furthermore, after step S830, the method may further include the following steps:

[0124] S831, obtain the area of ​​the Voronoi polygon corresponding to each centroid to obtain a list of Voronoi polygon areas Y = (Y1, Y2, ..., Y...). j , ..., Y m ), j=1,2,…,m; where, Y j Let m be the area of ​​the Voronoi polygon corresponding to the j-th centroid, and m be the number of centroids.

[0125] S832, if |Y j If -λ|>θ, then the j-th centroid is determined as an abnormal centroid; where λ is the average Voronoi polygon area corresponding to Y, and θ is the preset Voronoi polygon area difference threshold.

[0126] Objective judgment rules are constructed by statistical mean and threshold to avoid the subjectivity of manual inspection; the mean-based detection method can filter the interference of individual noise points, such as accidental small errors, and focus on anomalies that significantly deviate from the overall distribution; the abnormal centroid is directly associated with the specific glue point, which facilitates subsequent traceability of the production process, such as coating head blockage, mechanical vibration, etc.

[0127] S833, if the number of abnormal centroids is greater than the preset number threshold, it is determined that there is an abnormality in the sol-gel fabric at the point to be tested.

[0128] In this embodiment, if it is determined through steps S831-S833 that the sol-gel fabric at the point to be tested is abnormal, no further judgment is required, which can further improve the judgment efficiency.

[0129] S900: If there are no abnormalities in the sol-bonded fabric at the test point, hot pressing is performed to obtain the final fabric.

[0130] The manufacturing method of dot-bonded fabric in this embodiment divides the target image into sub-images containing complete glue dots, incomplete glue dots, or no glue dots according to the number of glue dots pixels and the image resolution. This avoids the redundant calculation of processing a massive number of pixels required by existing full-image traversal detection, focuses the detection range on the glue dot-related area, greatly reduces the amount of data processing, and significantly improves the detection efficiency, which can meet the stringent requirements of high-speed production lines for real-time detection.

[0131] Furthermore, by using DCT coefficient matrix feature clustering, and leveraging the differences in energy distribution in the frequency domain for different glue dot states, coefficient matrix clusters containing complete glue dots, incomplete glue dots, and glue-free dots are accurately separated. This effectively solves the problems of complete and incomplete glue dots and background noise aliasing in traditional methods, improving the purity and accuracy of glue dot feature extraction. Through energy center mapping and incomplete glue dot splicing, only the DCT coefficients corresponding to the glue dots are retained and mapped to the spatial domain to determine the centroid. At the same time, the segmented incomplete glue dot sub-images are spliced ​​and reconstructed, eliminating the interference of background noise on centroid localization and compensating for the localization deviation caused by glue dot incompleteness or segmentation, achieving high-precision localization of glue dot centroids. Based on multi-dimensional analysis of centroid distribution, combined with topological features and spatial pattern recognition, subtle changes in glue dot integrity, distribution uniformity, and positional accuracy can be sensitively captured, effectively identifying local anomalies and minor defects that are difficult to detect by traditional methods, significantly improving the sensitivity and reliability of detection.

[0132] Furthermore, although the steps of the method in this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.

[0133] Embodiments of the present invention also provide a non-transitory computer-readable storage medium that can be disposed in an electronic device to store at least one instruction or at least one program related to implementing a method in the method embodiments, wherein the at least one instruction or the at least one program is loaded and executed by the processor to implement the method provided in the above embodiments.

[0134] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0135] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting programs for use by or in conjunction with an instruction execution system, apparatus, or device.

[0136] The program code contained on the readable medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0137] Program code for performing the operations of this application can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0138] Embodiments of the present invention also provide an electronic device, including a processor and the aforementioned non-transitory computer-readable storage medium.

[0139] The electronic device is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments in this application.

[0140] Electronic devices are manifested in the form of general-purpose computing devices. Components of an electronic device may include, but are not limited to: at least one processor, at least one memory, and a bus connecting different system components (including memory and processor).

[0141] The memory stores program code that can be executed by the processor, causing the processor to perform the steps in the various embodiments described in this specification.

[0142] The memory may include readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory, and may further include read-only memory (ROM).

[0143] The memory may also include programs / utilities having a set (at least one) of program modules, including but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0144] A bus can represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus that uses any of the various bus structures.

[0145] The electronic device can also communicate with one or more external devices (e.g., keyboards, pointing devices, Bluetooth devices, etc.), one or more devices that enable user interaction with the electronic device, and / or any device that enables the electronic device to communicate with one or more other computing devices (e.g., routers, modems, etc.). This communication can be achieved through input / output (I / O) interfaces. Furthermore, the electronic device can communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter. The network adapter communicates with other modules of the electronic device via a bus. Other hardware and / or software modules can be used in conjunction with the electronic device, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0146] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0147] Embodiments of the present invention also provide a computer program product including program code, which, when the program product is run on an electronic device, causes the electronic device to perform the steps of the methods described above in various exemplary embodiments of the present invention.

[0148] While specific embodiments of the invention have been described in detail by way of examples, those skilled in the art should understand that the examples are for illustrative purposes only and are not intended to limit the scope of the invention. Those skilled in the art should also understand that various modifications can be made to the embodiments without departing from the scope and spirit of the invention.

Claims

1. A method for manufacturing a dot-sol fabric, characterized in that, The method includes the following steps: S100, acquire the target image corresponding to the sol-gel fabric point to be detected; the target image contains several glue points; S200, based on the number of pixels corresponding to each glue dot in the target image and the resolution of the target image, divide the target image into several sub-images; among these sub-images, there are sub-images that contain only one complete glue dot; S300, perform DCT transformation on each sub-image to obtain the DCT coefficient matrix corresponding to each sub-image; S400, based on the characteristics of each DCT coefficient matrix, cluster all DCT coefficient matrices to obtain a cluster QA of DCT coefficient matrices containing a complete glue dot; S500, process each DCT coefficient matrix in QA to obtain the processed DCT coefficient matrix cluster WA corresponding to QA; wherein, each DCT coefficient matrix in WA contains only the DCT coefficients corresponding to the glue dot image; S600 maps the energy center corresponding to each DCT coefficient matrix in WA to the target image to obtain the centroid of the glue point corresponding to each DCT coefficient matrix in QA. S700, stitch together the sub-images corresponding to each DCT coefficient matrix that is not in QA in the target image to obtain the centroid of the complete glue dot composed of incomplete glue dots; S800 determines whether there are any abnormalities in the sol-gel fabric at the point to be detected based on the centroid distribution of each glue dot in the target image. S900: If there are no abnormalities in the sol-bonded fabric at the test point, hot pressing is performed to obtain the final fabric.

2. The method for manufacturing the dot-sol fabric according to claim 1, characterized in that, Step S200 includes the following steps: S210, obtain the longest side of the smallest bounding box corresponding to each glue dot in the target image, to obtain a list of longest sides H = (H1, H2, ..., H...). i H n ), i = 1, 2, ..., n; where H i Let n be the longest side of the bounding box corresponding to the i-th glue dot, and n be the number of glue dots. S220, obtain the target longest side MH=MAX(H); MAX() is the preset maximum value function; S230, obtain the number of pixels NUM1 corresponding to MH; S240, if NUM1=2 r Therefore, the resolution of the sub-image is determined to be 2. r ×2 r ; r is an integer greater than 0; S250, if 2 r <NUM1<2 r+1 Therefore, the resolution of the sub-image is determined to be 2. r+1 ×2 r+1 .

3. The method for manufacturing the dot-sol fabric according to claim 1, characterized in that, Step S500 includes the following steps: S510, Obtain the standard DCT coefficient matrix D of the non-sol-bonded fabric corresponding to QA. bg ; where D bg The images were obtained by performing DCT transformation on the images of the non-sol fabric. The surface tension, material and knitting method of the fabric were the same during the process of photographing the non-sol fabric and the fabric to be tested for the sol. S520, for any DCT coefficient matrix D in QA fg D fg With D bg Perform subtraction on the same digit to obtain D. fg With D bg The corresponding DCT coefficient difference matrix D diff =(DE1, DE2, ..., DE p , ..., DE q ), p=1,2,…,q; where DE p D fg With D bg The difference corresponding to the p-th coefficient, q is D fg With D bg The number of coefficients in the middle; S530, Traverse D diff For each coefficient in the equation, if -η < DE p If <η, then D fg Set the p-th coefficient in the matrix to 0 to obtain D. fg The corresponding processed DCT coefficient matrix; η is the preset DCT coefficient difference threshold.

4. The method for manufacturing the dot-sol fabric according to claim 1, characterized in that, Step S600 includes the following steps: S610, for any DCT coefficient matrix D in WA e , obtain D e Corresponding total energy Where N-1 is D e The number of rows or columns; (u, v) is D e The frequency coordinates; E(u, v) is the energy of the coefficient corresponding to (u, v); E(u, v) = |D e (u, v) | 2 ;D e (u, v) are the coefficients corresponding to (u, v); S620, based on E(u, v) and E total Determine D e Frequency coordinates of the energy center (u) center v center ); ; ; S630, will (u center v center ) converted to D e The corresponding spatial coordinates within the sub-image are used to obtain D. e The centroid of the corresponding glue dot.

5. The method for manufacturing the dot-sol fabric according to claim 1, characterized in that, Step S700 includes the following steps: S710, stitch together the sub-images corresponding to each DCT coefficient matrix that is not in QA in the target image to obtain a complete glue dot composed of several incomplete glue dots; S720, obtain the outline of each complete glue dot; S730 determines the centroid of each complete glue dot based on its outline.

6. The method for manufacturing the dot-sol fabric according to claim 5, characterized in that, Step S800 includes the following steps: S810 converts the centroid of each glue dot into a two-dimensional spatial point set and constructs the Delaunay triangulation and Voronoi diagram corresponding to the two-dimensional spatial point set; S820, Obtain the topological features corresponding to the Delaunay triangulation; the topological features include: the number of adjacent nodes of each centroid, the average side length of the Delaunay triangulation, the standard deviation of the side length, and the distribution of the interior angles of the triangles; S830, Based on the Voronoi diagram, extract the Voronoi polygon features corresponding to each centroid; the Voronoi polygon features include: the area, perimeter, and compactness of the polygon; S840 generates the glue dot feature vector XL corresponding to the target image based on the topological features corresponding to the Delaunay triangulation and the Voronoi polygon features corresponding to each centroid. S850, if the similarity between XL and the preset standard glue dot feature vector is greater than the preset similarity threshold, then it is determined that there is no abnormality in the sol-gel fabric at the point to be detected.

7. The method for manufacturing the dot-sol fabric according to claim 6, characterized in that, Following step S830, the method further includes the following steps: S831, obtain the area of ​​the Voronoi polygon corresponding to each centroid to obtain a list of Voronoi polygon areas Y = (Y1, Y2, ..., Y...). j , ..., Y m ), j=1,2,…,m; where, Y j Let m be the area of ​​the Voronoi polygon corresponding to the j-th centroid, and m be the number of centroids. S832, if |Y j If -λ|>θ, then the j-th centroid is determined as an abnormal centroid; where λ is the average Voronoi polygon area corresponding to Y, and θ is the preset Voronoi polygon area difference threshold. S833, if the number of abnormal centroids is greater than the preset number threshold, it is determined that there is an abnormality in the sol-gel fabric at the point to be tested.

8. A non-transitory computer-readable storage medium, wherein the storage medium stores at least one instruction or at least one program segment, characterized in that, The at least one instruction or the at least one program segment is loaded and executed by the processor to implement the method for manufacturing the dot-sol fabric as described in any one of claims 1-7.

9. An electronic device, characterized in that, Includes a processor and the non-transitory computer-readable storage medium as described in claim 8.

Images