Textile and garment defect pattern data processing method based on defect judgment model

Abstract

The application discloses a textile and garment flaw pattern data processing method based on a flaw judgment model, and relates to the technical field of data processing.The method comprises the following steps: rough screening, fine screening and comprehensive decision-making.The application firstly judges whether each textile and garment pattern has structural flaws, if the textile and garment pattern has structural flaws, structural flaw feedback correction is performed, otherwise, texture data and color data of each textile and garment pattern are obtained, whether the texture data and the color data have texture flaws and color flaws is respectively judged, if the texture data and the color data have texture flaws and color flaws, the severity of the flaws is evaluated based on the texture data and the color data to determine whether to perform rework repair, otherwise, the textile and garment pattern is qualified, the quality judgment accuracy of the textile and garment pattern is improved, and the problem that the quality judgment accuracy of the textile and garment pattern is low due to the inaccurate identification of the flaws of the textile and garment pattern in the prior art is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a textile and garment defect pattern data processing method based on a defect judgment model. BACKGROUND

[0002] In the textile and garment industry, consumers' quality requirements are constantly rising, production efficiency needs to be improved, and intelligent transformation is imminent. In the current competitive and rapidly developing textile and garment industry, consumers' quality awareness is constantly awakening, and the quality requirements for textile and garment products are showing a trend of continuous rise. No longer just satisfied with the basic style and function of clothing, consumers are increasingly concerned about whether there are defects on the surface of the product, such as tiny holes, difficult-to-detect color differences, and texture uniformity. Extremely tiny defects can cause consumers' dissatisfaction, leading to an increase in return rate and causing serious damage to brand image.

[0003] The current technical implementation process is image acquisition, data preprocessing, feature extraction / enhancement, data labeling, model training / reasoning, post-processing, and result output / rework suggestion. Image acquisition means scanning the fabric by an industrial camera (linear array / surface array); data preprocessing mainly includes image cropping / blocking (commonly 256x256, 512x512), denoising / smoothing / gray scale conversion, and light balance (such as CLAHE contrast enhancement, Contrast Limited Adaptive Histogram Equalization); feature extraction / enhancement includes random flipping, rotation, scaling, artificial defect synthesis (simulation of broken weft, stains), and multi-scale fusion (adaptation to small defects); data labeling includes manual labeling (Labelme, CVAT, Computer Vision Annotation Tool) and semi-automatic labeling (edge detection + manual revision); common model architectures for model training / reasoning include but are not limited to ResNet (Residual Network), EfficientNet (Efficient Network), YOLOv5 / v8, Faster-RCNN (Faster Region Convolutional Neural Network), U-Net (U-shaped Network), Mask R-CNN U-Net, Mask R-CNN, Autoencoder (Autoencoder), GAN (Generative Adversarial Network), automatic encoder, and GAN; post-processing includes non-maximum suppression (NMS, Non-Maximum Suppression) deduplication and sliding window splicing / heat map fusion; result output / rework suggestion outputs image labeling results (superimposed defect frame / mask) and data information (type, location, area, severity).

[0004] For example, the Chinese invention patent with publication number: CN116611503B discloses a lightweight model construction method and device for multi-category defect real-time detection, which includes: obtaining a target object image containing multi-category defects to form an image data set; building an initial lightweight model and setting the hyperparameters of the lightweight model training, wherein the initial lightweight model includes a lightweight backbone network, a neck network and a head detection network, and the lightweight backbone network is used to extract multi-scale image features fused with position information; model training iteration is performed on the initial lightweight model according to the image data set until convergence; and a lightweight model for multi-category defect real-time detection is constructed.

[0005] For example, the Chinese invention patent with publication number: CN112036541B discloses a fabric defect detection method based on genetic algorithm optimized neural network, which includes: initializing Gabor filter parameters; collecting fabric defect images; marking the defect categories and the bounding boxes containing defects to establish a PascalVOC data set; sending the PascalVOC data set into a Faster-RCNN network training model for training and calculating mAP; taking mAP as the fitness function of the genetic algorithm, and performing mutation, crossover and selection to obtain offspring Gabor parameters until the number of iterations reaches the set maximum value, and outputting the optimal genotype, i.e. the optimal filter parameters of the fabric defect image; calling the corresponding Faster-RCNN model to output the position, type and accuracy of the fabric defect.

[0006] The above-mentioned technology at least has the following technical problems:

[0007] In the prior art, a single detection process is usually used to detect all patterns to the same extent, which lacks pertinence. It cannot quickly screen out structure defects that seriously affect quality for priority processing, nor can it reasonably allocate computing resources, resulting in low detection efficiency. For textile and garment enterprises with large-scale production, it is difficult to meet the needs of efficient production, making it difficult to effectively guarantee product quality, and the comprehensive detection accuracy is usually low. In addition, there is a lack of scientific quantitative evaluation method for the severity of defects in the key link of determining whether the product needs to be reworked, so that a large number of products that can meet the standards through simple repair are misjudged as scrap, causing unnecessary waste of resources such as raw materials, manpower and time; at the same time, some products that have serious defects and should not flow into the market are released due to judgment errors, which seriously affects the stability of product quality, damages the reputation of enterprises and the rights and interests of consumers, and there is a problem of inaccurate identification of textile and garment pattern defects leading to low accuracy of textile and garment pattern quality judgment. SUMMARY

[0008] The embodiment of the application provides a textile and garment defect pattern data processing method based on a defect judgment model, and solves the problem of low accuracy of textile and garment pattern quality judgment caused by inaccurate identification of textile and garment pattern defects, and improves the accuracy of textile and garment pattern quality judgment.

[0009] The embodiment of the application provides a textile and garment defect pattern data processing method based on a defect judgment model, and solves the problem of low accuracy of textile and garment pattern quality judgment caused by inaccurate identification of textile and garment pattern defects, and improves the accuracy of textile and garment pattern quality judgment.

[0010] The one or more technical solutions provided in the embodiment of the application have at least the following technical effects or advantages:

[0011] 1. The structure class data of each textile and garment pattern is obtained to determine whether there is a structure class defect, if there is a structure class defect, structure class defect feedback correction is performed, in the structure class defect detection, the structure problem that seriously affects the quality of the garment can be quickly located, and the structural defect of the textile and garment pattern is determined preferentially, thereby realizing rough screening of the textile and garment pattern, improving the efficiency of screening the textile and garment pattern, otherwise, the texture class data and color class data of each textile and garment pattern are obtained, and whether there is a texture class defect and a color class defect is determined, for the pattern without a structure class defect, the texture and color class defects are further detected in depth, and the gradually deepening detection process improves the detection efficiency, avoids complex detection of all patterns, saves the calculation resources and time cost, and ensures the comprehensiveness of the detection, if there is a texture class defect or a color class defect, the severity of the defect is evaluated based on the texture class data and the color class data of each textile and garment pattern to determine whether each textile and garment pattern is repaired, otherwise, the textile and garment pattern is qualified, thereby realizing comprehensive quantification of the severity of the defect of each textile and garment pattern, improving the scientificity and rationality of the decision, more effectively guiding the production, reducing the cost waste caused by inaccurate defect judgment, and improving the accuracy of textile and garment pattern quality judgment, effectively solving the problem of low accuracy of textile and garment pattern quality judgment caused by inaccurate identification of textile and garment pattern defects.

[0012] 2、By extracting the structure class data of each textile and garment pattern and inputting it into the textile and garment pattern structure class defect judgment model of the textile and garment pattern, and outputting the corresponding structure class defect judgment data, the textile and garment pattern is roughly screened, the efficiency of structure class defect judgment is improved, if the textile and garment pattern structure class defect judgment model outputs the structure class defect first judgment value, it is judged that the corresponding textile and garment pattern has structure class defect and structure class defect feedback correction is carried out, so as to improve the accuracy of textile and garment pattern structure class defect judgment, otherwise it is judged that the corresponding textile and garment pattern does not have structure class defect, so as to quickly locate the structure problem that seriously affects the quality of the garment, and then realize the rapid screening of the structure defect of each textile and garment pattern.

[0013] 3、By extracting the texture class data of each textile and garment pattern with texture class defect, and combining the corresponding texture class parameter threshold for relative deviation processing, a solid data foundation is laid, and based on the correlation between the processing result and the texture defect severity coefficient, coupling operation is carried out to obtain the texture defect severity coefficient, which quantifies the severity of the texture class defect of the corresponding textile and garment pattern. At the same time, the color class data of each textile and garment pattern with color class defect is extracted, and the relative deviation operation is carried out combined with the corresponding color class parameter threshold, which preliminarily quantifies the difference degree of the color class data and the corresponding color class parameter threshold. According to the correlation between the operation result and the color defect severity coefficient, coupling operation is carried out to obtain the color defect severity coefficient, which quantifies the severity of the color class defect of the corresponding textile and garment pattern. Finally, the sum operation of the texture defect severity coefficient of each textile and garment pattern is carried out to obtain the defect severity coefficient of each textile and garment pattern, and the two kinds of coefficients are combined to obtain the defect severity coefficient of each textile and garment pattern. Compared with the threshold value in the database, it can accurately judge whether the product needs to be reworked, and realize the fine decision of each textile and garment pattern. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 The flow chart of the textile and garment defect pattern data processing method based on the defect judgment model provided by the embodiment of the application;

[0015] Figure 2 The flow chart of the modified classification provided by the embodiment of the application;

[0016] Figure 3 The flow chart of the judgment of the texture class defect and the color class defect of each textile and garment pattern provided by the embodiment of the application. DETAILED DESCRIPTION

[0017] The embodiment of the present application provides a textile and garment flaw pattern data processing method based on a flaw judgment model, solves the problem of low accuracy of quality judgment of textile and garment patterns caused by inaccurate flaw identification of textile and garment patterns, and has the following advantages: the structural class data of each textile and garment pattern is input into a textile and garment pattern structural class flaw judgment model of the textile and garment pattern, and corresponding structural class flaw judgment data is output, if the textile and garment pattern structural class flaw judgment model outputs a structural class flaw first judgment value, it is judged that the corresponding textile and garment pattern has a structural class flaw and structural class flaw feedback correction is performed, otherwise, it is judged that the corresponding textile and garment pattern does not have a structural class flaw, and texture class data and color class data of each textile and garment pattern are acquired, whether there is a texture class flaw and a color class flaw is judged respectively, if there is a texture class flaw or a color class flaw, the flaw severity is evaluated based on the texture class data and the color class data of each textile and garment pattern to judge whether each textile and garment pattern needs to be repaired, otherwise, the textile and garment pattern is qualified, and the accuracy of quality judgment of textile and garment patterns is improved.

[0018] The technical solution in the embodiment of the present application solves the problem of low accuracy of quality judgment of textile and garment patterns caused by inaccurate flaw identification of textile and garment patterns, and the general idea is as follows:

[0019] By judging whether each textile and garment pattern has a structural class flaw, if the textile and garment pattern has a structural class flaw, structural class flaw feedback correction is performed, otherwise, texture class data and color class data of each textile and garment pattern are acquired, whether there is a texture class flaw and a color class flaw is judged respectively, if there is a texture class flaw or a color class flaw, the flaw severity is evaluated based on the texture class data and the color class data to judge whether each textile and garment pattern needs to be repaired, otherwise, the textile and garment pattern is qualified, and the accuracy of quality judgment of textile and garment patterns is improved.

[0020] In order to better understand the above technical solution, the above technical solution will be described in detail in combination with the drawings of the specification and specific embodiments.

[0021] As shown in Figure 1 Fig. 1 is a flowchart of a textile and garment flaw pattern data processing method based on a flaw judgment model provided by the embodiment of the present application, and the method comprises the following steps:

[0022] S1, acquire the structural class data of each textile and garment pattern to judge whether each textile and garment pattern has a structural class flaw, if each textile and garment pattern has a structural class flaw, perform structural class flaw feedback correction, otherwise, perform S2.

[0023] S2, acquire the texture data and color data of each textile garment pattern, respectively determine whether each textile garment pattern has texture defects and color defects, if each textile garment pattern has texture defects or color defects, then execute S3, otherwise, record the textile garment pattern as qualified.

[0024] S3, based on the texture data and color data of each textile garment pattern, evaluate the defect severity of each textile garment pattern to determine whether to repair each textile garment pattern.

[0025] In this embodiment, in order to solve the problem that the identification of defects of textile garment patterns is not accurate, resulting in low accuracy of textile garment pattern quality determination, the present application adopts a staged detection method of first roughly screening structural defects and then finely screening texture and color defects, realizes rapid positioning of structural problems that seriously affect garment quality, and for patterns without structural defects, further detects texture and color defects, the gradually deepening detection process not only improves the detection efficiency, but also avoids complex detection of all patterns; in addition, the automatic determination method also reduces the interference of human factors, improves the scientificity and rationality of decision-making, can more effectively guide production, improves the accuracy of textile garment pattern quality determination, and reduces the cost waste caused by inaccurate defect judgment.

[0026] Before designing the textile garment defect pattern data processing method based on the defect judgment model, professional technicians will first construct a preset database for supporting the operation of various methods. The database contains multiple key parameter information, such as: fabric flatness preset data, structure classification threshold, structure parameter threshold, texture parameter threshold, color parameter threshold, defect severity coefficient threshold, preset rework times, and preset proportion, etc. These parameters are all preset by technicians with professional background according to the specific method requirements and hardware resource conditions. And the preset database provides key data support for subsequent data uploading, storage optimization, screening judgment and other automatic processing processes.

[0027] Further, whether each textile garment pattern has structural defects is determined, and the specific process is as follows:

[0028] First, extract the structural data of each textile garment pattern, establish a multi-dimensional data input system for structural defects of textile garment patterns, cover the core indicators of fabric quality, and the structural data includes hole area, yarn length and warp and weft density.

[0029] In one specific embodiment, the structural class data of each textile garment pattern is obtained, and the specific implementation method is as follows: a line scan camera is used to scan each textile garment pattern, high-resolution images of each textile garment pattern are collected, the collected images of each textile garment pattern are transmitted to a computer for preprocessing, the preprocessed images of each textile garment pattern are subjected to binaryzation processing, based on a connected domain analysis method of a binary image, the hole area of each textile garment pattern is obtained, the preprocessed images of each textile garment pattern are subjected to morphological operation, the yarn length of each textile garment pattern is obtained, and the preprocessed images of each textile garment pattern are subjected to Fourier transform, and the warp and weft density of each textile garment pattern is obtained. The line scan camera (Line Scan Camera) is an image acquisition device, which is different from a traditional area array camera, adopts a line-by-line scanning mode to capture images, is different from the area array camera which captures the whole image at the same time, scans only one horizontal line of the image each time, and then captures continuous image lines by moving the object or the camera. It is usually used in scenarios requiring continuous and accurate scanning, such as production lines, fabric detection, printing quality control, etc.

[0030] It should be noted that the hole area of each textile garment pattern is obtained by the following method: the preprocessed gray-scale image is converted into a black-and-white image by a fixed threshold (such as 127) or an adaptive threshold (such as the Otsu algorithm), the hole area is highlighted as a white pixel, and the background is a black pixel. An 8-neighbor connectedness (up, down, left, right and diagonal adjacent pixels are considered as connected) is used to traverse the binary image, and all independent white areas are marked. The total number of pixels in each connected area is the area of the connected area, and thus the hole area of each textile garment pattern is obtained. The yarn length of each textile garment pattern is obtained by the following method: a thinning algorithm is applied to the binary image, the yarn area is eroded into a single-pixel-width continuous skeleton, the arc length of the contour is calculated, and the arc length is taken as the yarn length of each textile garment pattern. The warp and weft density of each textile garment pattern is obtained by the following method: a two-dimensional fast Fourier transform is performed on the gray-scale image, the image is converted from the spatial domain to the frequency domain, periodic peaks are found in the frequency spectrum, the periodic peaks correspond to the arrangement frequency of the warp and weft yarns in the image, the frequency corresponding to the horizontal direction peak value and the frequency corresponding to the vertical direction peak value are calculated, and thus the warp and weft density of each textile garment pattern is obtained.

[0031] In the second step, the structural class data is input into the textile garment pattern structure class defect judgment model of the textile garment pattern to improve the recognition efficiency and consistency, and the structural class defect judgment data of each textile garment pattern is output. The structural class defect judgment data includes a structural class defect first judgment value and a structural class defect second judgment value. It should be noted that in the textile garment pattern structure class defect judgment model, the structural class defect first judgment value is 1, and the structural class defect second judgment value is 0, which is convenient for subsequent automatic processing and also reserves space for expansion to multiple levels of classification.

[0032] Thirdly, if the structure defect judgment model outputs the first judgment value of the structure defect, it is determined that the corresponding textile garment pattern has a structure defect and a structure defect feedback correction is performed, and if the structure defect judgment model outputs the second judgment value of the structure defect, it is determined that the corresponding textile garment pattern does not have a structure defect, thereby realizing automatic recognition and decision-driven processing of the structure defect and improving the degree of automation of the production line.

[0033] In this embodiment, by extracting the structure data of each textile garment pattern, it is helpful for the subsequent model to accurately identify different types of structural defects, thereby improving the recognition accuracy and adaptability of the textile garment pattern. Meanwhile, the structure data is input into the structure defect judgment model of the textile garment pattern, and the corresponding structure defect judgment data is output, thereby realizing an intelligent conversion process from data to judgment result, reducing manual intervention, improving detection efficiency, and determining whether there is a structure defect according to the structure defect judgment data. This not only avoids the false positive probability, but also improves the processing efficiency and production resource utilization rate.

[0034] Further, the structure data of each textile garment pattern is obtained, which also includes textile garment fabric flatness determination, and the specific steps are as follows:

[0035] Step one, obtain the fabric flatness parameter, which includes the gray value gradient, the local direction feature entropy value and the edge curvature.

[0036] It should be noted that the gray value gradient is usually obtained by Sobel operator or Prewitt, Scharr and the like to obtain the horizontal direction gray value and the vertical direction gray value, and then the gradient amplitude is calculated based on the horizontal direction gray value and the vertical direction gray value to obtain the gray value gradient. First, the image window is divided, then the gradient response of each direction (such as 0°, 45°, 90°, 135°) is calculated, and the entropy value is calculated after normalization to obtain the local direction feature entropy value. First, the edge points are obtained by Canny, then the edge point coordinates are extracted, and the finite difference method is used to approximate the first and second derivatives of the edge point coordinates, and the curvature formula is substituted to calculate the edge curvature.

[0037] Step two, based on the fabric flatness parameter and the fabric flatness preset data obtained from the preset database, the judgment is performed, a standardized and parameterized quality evaluation system is established, the objectivity and repeatability of the detection are improved, and the fabric flatness preset data includes the gray value gradient preset value, the local direction feature entropy value preset value and the edge curvature preset value. Among them, the fabric flatness preset data is input into the preset database by the preset production personnel in advance for storage, and can be directly extracted from the preset database when used.

[0038] Step three, if the fabric flatness parameter is lower than the fabric flatness preset data, prompt the preset staff to adjust the flatness of the textile fabric, otherwise feedback the textile fabric flatness is qualified, realize real-time quality control and feedback mechanism, timely troubleshoot quality risk source.

[0039] In this embodiment, based on the comparison between the fabric flatness parameter and the corresponding fabric flatness preset data, not only the accuracy of scanning the textile garment pattern is ensured, but also the positioning accuracy of subsequent structural defects is prevented from being affected; and through flatness adjustment and feedback, the uneven fabric entering the subsequent textile garment pattern defect detection or sewing process is reduced, avoiding quality transmission and resource waste.

[0040] Further, the structural defect feedback correction also includes correction classification, the specific steps are as follows:

[0041] C1, based on the structure data, respectively compare with the structure classification threshold value stored in the preset database, the structure classification threshold value includes the hole area accumulation threshold value, the thread length accumulation threshold value and the warp and weft density accumulation threshold value; wherein, the structure classification threshold value is pre-input by the preset production personnel into the preset database for storage, and can be directly extracted from the preset database when used; through the comparison between the above structure data and the structure classification threshold value, automatic defect judgment is realized, avoiding manual misjudgment.

[0042] C2, if the hole area of each textile garment pattern is greater than the hole area accumulation threshold value, the corresponding textile garment pattern is transmitted to the hole scrap conveyor belt, avoiding the flow of serious holes into the subsequent link, improving the product quality stability, otherwise it is transmitted to the hole repair belt and executes C3, if the number of textile garment patterns transmitted to the hole scrap conveyor belt exceeds the preset hole accumulation number, the preset production personnel is prompted to detect the textile fabric to detect whether the fabric has quality problems, through the accumulation number early warning mechanism, not only the automatic identification of the quality fluctuation trend of the textile garment pattern is realized, but also the batch material problem is found more quickly.

[0043] C3, if the thread length of each textile garment pattern is greater than the thread length accumulation threshold value, the corresponding textile garment pattern is transmitted to the thread scrap conveyor belt, otherwise it is transmitted to the thread repair belt and executes C4, realizing the fine-grained judgment of thread defects, distinguishing between repairable and non-repairable, if the number of textile garment patterns transmitted to the thread scrap conveyor belt exceeds the preset thread accumulation number, the corresponding textile garment pattern is marked as defective, through the automatic marking of defective when the thread defect accumulation exceeds the standard, not only the product quality control efficiency of the textile garment pattern is improved, but also the frequency of manual intervention is reduced.

[0044] C4, if the warp and weft density of each textile garment pattern is greater than the warp and weft density accumulation threshold, the corresponding textile garment pattern is transmitted to the density scrap conveyor belt, otherwise it is transmitted to the density repair belt, if the number of textile garment patterns transmitted to the density scrap conveyor belt exceeds the preset density accumulation number, the preset production personnel is prompted to detect whether the process parameters are correct, the process parameters include weaving tension and weft feeding speed; Through the above indirect monitoring of process parameters for the abnormal accumulation of warp and weft density, early warning of equipment abnormalities is realized.

[0045] In the embodiment, by combining the above structure data analysis with the comparison of the corresponding structure classification threshold, not only the standardization degree of textile garment pattern quality detection is improved, and the waste of repair resources is reduced through repair classification, the efficiency of flaw processing resource is improved, but also a high-precision, multi-level, traceable flaw detection and response mechanism is realized, which has good automation and industrial adaptability, and improves the response ability of the system to the density quality problems of the textile garment pattern. The closed-loop reminding mechanism makes the equipment adjustment more timely.

[0046] As shown in Figure 2 The flowchart of the modified classification provided by the embodiment of the application is shown in the figure, and the specific logic is: compare the structure data with the structure classification threshold, the structure classification threshold includes the hole area accumulation threshold, the thread length accumulation threshold and the warp and weft density accumulation threshold; if the hole area of each textile garment pattern is greater than the hole area accumulation threshold, it is transmitted to the hole scrap conveyor belt, otherwise it is transmitted to the hole repair belt and C3 is executed, if the number of textile garment patterns transmitted to the hole scrap conveyor belt exceeds the hole accumulation number, the preset production personnel is prompted to detect the textile garment fabric; C3, if the thread length of each textile garment pattern is greater than the thread length accumulation threshold, it is transmitted to the thread scrap conveyor belt, otherwise it is transmitted to the thread repair belt and C4 is executed, if the number of textile garment patterns transmitted to the thread scrap conveyor belt exceeds the thread accumulation number, the corresponding textile garment pattern is marked as defective; C4, if the warp and weft density of each textile garment pattern is greater than the warp and weft density accumulation threshold, it is transmitted to the density scrap conveyor belt, otherwise it is transmitted to the density repair belt, if the number of textile garment patterns transmitted to the density scrap conveyor belt exceeds the density accumulation number, the preset production personnel is prompted to detect whether the process parameters are correct; Through the above process, the consistency of textile garment pattern quality detection is improved, and the refinement of textile garment pattern repair is improved.

[0047] Further, it is judged whether each textile garment pattern has texture class defects and color class defects respectively, and the specific steps are as follows:

[0048] P1, extract the texture class data of each textile garment pattern, the texture class data includes the final contrast, the final entropy value and the roughness.

[0049] In one specific embodiment, the texture class related data of each textile garment pattern is obtained, and the specific implementation method is as follows: the preprocessed image of each textile garment pattern is extracted, a gray level co-occurrence matrix is constructed, the number of occurrences of each direction pixel and the gray level of the neighborhood pixel pair with a specified direction and step length is counted, the probability matrix of each direction is normalized, the contrast value of each direction is calculated through the contrast calculation formula, the mean value is processed to obtain the final contrast of each textile garment pattern, the probability matrix of each direction is extracted, the entropy of each direction is calculated through the entropy calculation formula, the mean value is processed to obtain the final entropy value of each textile garment pattern, the preprocessed image of each textile garment pattern is extracted, the fractal dimension is calculated by box counting method to obtain the roughness of each textile garment pattern.

[0050] It should be noted that the probability matrix , wherein (p, q) is a pixel pair, S(p, q) is the number of occurrences of the pixel pair, S' is the total number of similar pairs, the contrast calculation formula is: , the entropy calculation formula is: , the greater the entropy value, the more complex and random the texture, and the box counting method calculates the fractal dimension, which is a prior art.

[0051] P2, input into the textile garment pattern texture class defect judgment model for automatic judgment, realizing an intelligent and standardized discrimination process, and outputting the texture class defect judgment data of each textile garment pattern, which includes the texture class defect first judgment value and the texture class defect second judgment value.

[0052] P3, if the textile garment pattern texture class defect judgment model outputs the texture class defect first judgment value, it is determined that the corresponding textile garment pattern has texture class defects, and if the textile garment pattern texture class defect judgment model outputs the texture class defect second judgment value, it is determined that the corresponding textile garment pattern does not have texture class defects, and once the texture class defect is determined, the back-end production management system can be automatically marked as defective, controlled to be offline or classified into storage, thereby improving the overall quality consistency.

[0053] P4, determining whether each textile garment pattern has color class defects.

[0054] Through the above process, not only the textile garment pattern texture class defect judgment model is applied for automatic judgment, but also an intelligent and standardized discrimination process is realized, and texture class defect judgment is performed according to the texture class defect judgment data, thereby avoiding the flow of textile garment patterns with texture blur or roughness problems into the finished product link, and improving the overall quality consistency.

[0055] It should be noted that whether each textile garment pattern has color class defects is determined, and the specific process is as follows:

[0056] Firstly, color class data of each textile garment pattern is extracted, and a digital measurement standard of color information is established, which is conducive to realizing quantifiable, analyzable and comparable color quality. The color class data includes color difference and color histogram difference value.

[0057] In one specific embodiment, color class related data of each textile garment pattern is obtained. Specifically, the following steps are implemented: the preprocessed image of each textile garment pattern is extracted, the image of each textile garment pattern is converted into CIELAB (Commission Internationale de l'Éclairage L*a*b*) color space through color space conversion technology, the color difference of each pixel point is calculated through a color difference calculation formula, the mean value is processed to obtain the color difference value of each textile garment pattern, and the preprocessed image of each textile garment pattern is compared with the standard image of the textile garment pattern stored in the database to calculate the color histogram difference value of each textile garment pattern and the standard image.

[0058] It should be noted that the color difference wherein, ∆L, ∆a, ∆b represent the difference values of brightness, red-green component and yellow-blue component respectively, the color histogram difference value of each textile garment pattern and the standard image is calculated, i.e. the Bhattacharyya distance, which is a prior art. The Bhattacharyya distance is used to measure the similarity of two probability distributions. In color histogram comparison, it can be used to measure the similarity of color distribution of two images. The value closer to 0 indicates that the two distributions are more similar, and the value closer to 1 indicates that the two distributions are more different.

[0059] Then, the color class data is input into the textile garment pattern color class defect judgment model to improve the accuracy and consistency of the output result, and the color class defect judgment data of each textile garment pattern is output. The color class defect judgment data includes color class defect first judgment value and color class defect second judgment value.

[0060] Finally, if the textile garment pattern color class defect judgment model outputs the color class defect first judgment value, it is determined that the corresponding textile garment pattern has color class defects. An explicit judgment standard is established, so that the system can automatically respond and classify the pattern samples. If the textile garment pattern color class defect judgment model outputs the color class defect second judgment value, it is determined that the corresponding textile garment pattern does not have color class defects.

[0061] In this embodiment, through the joint modeling judgment of pattern texture class and color class data, multi-dimensional fine detection of the appearance quality of textile and garment patterns is realized, the fusion extraction and intelligent model judgment of texture data have high fault tolerance and strong adaptability, and effectively reduce the deviation of manual judgment; at the same time, by constructing a color class data index system (color difference, histogram difference), quantitative and automatic defect identification of color information of textile and garment patterns is also realized, and the accuracy of identifying defects of textile and garment patterns is improved.

[0062] As shown in Figure 3 The flowchart for judging the texture class defects and color class defects of each textile and garment pattern provided by the embodiment of the application is shown in the figure, and the specific logic is as follows: whether each textile and garment pattern has texture class defects and color class defects is judged respectively, and the specific steps are as follows: the texture class data of each textile and garment pattern is extracted and input into the textile and garment pattern texture class defect judgment model, and the texture class defect judgment data of each textile and garment pattern is output, if the first judgment value of the texture class defect is output by the textile and garment pattern texture class defect judgment model, it is judged that the corresponding textile and garment pattern has texture class defects, otherwise it is judged that the corresponding textile and garment pattern does not have texture class defects; whether each textile and garment pattern has color class defects is judged, and the specific process is as follows: the color class data of each textile and garment pattern is extracted and input into the textile and garment pattern color class defect judgment model, and the color class defect judgment data of each textile and garment pattern is output, if the first judgment value of the color class defect is output by the textile and garment pattern color class defect judgment model, it is judged that the corresponding textile and garment pattern has color class defects, otherwise it is judged that the corresponding textile and garment pattern does not have color class defects; through the above process, the defect judgment of the textile and garment pattern is further refined, and the efficiency of the quality judgment of the textile and garment pattern is improved.

[0063] As a further explanation of the defect judgment data acquisition, it specifically includes three categories:

[0064] The first category is the specific acquisition method of the structure class defect judgment data as follows: the pre-defined structure class parameter threshold value is obtained from the database, the structure class parameter threshold value includes the hole area threshold value, the filament length threshold value and the warp and weft density threshold value; the structure class data is subjected to structure condition judgment with the structure class parameter threshold value to obtain the corresponding structure class defect judgment data, realizing efficient identification of structure class defects such as broken weft, sparse, hole and filament, improving the overall physical structure quality control capability of the fabric surface, and the structure condition judgment represents whether the structure class data is not greater than the structure class parameter threshold value.

[0065] It should be noted that the hole area threshold value, the filament length threshold value and the warp and weft density threshold value of the textile and garment pattern pre-defined in the database are set by professionals.

[0066] The specific expression of the textile and garment pattern structure class defect judgment model is as follows:

[0067] ;

[0068] In the formula, ai represents the judgment result of whether the i-th textile and garment pattern has a structure class defect, Ai represents the hole area of the i-th textile and garment pattern, Bi represents the length of the i-th textile and garment pattern, Ci represents the warp and weft density of the i-th textile and garment pattern, A' represents a hole area threshold value, B' represents a length threshold value, C' represents a warp and weft density threshold value, i represents the number of the textile and garment pattern, and i = 1, 2, …, j, j being a positive integer greater than 2.

[0069] The second type is the texture class defect judgment data, and the specific acquisition method is as follows: obtaining a predefined texture class parameter threshold value from the database, the texture class parameter threshold value including a final contrast threshold value, a final entropy threshold value and a roughness threshold value; and performing texture condition determination on the texture class data and the texture class parameter threshold value to obtain corresponding texture class defect judgment data, thereby realizing identification of defects such as pattern blur, local fuzzing, yarn shadowing and the like, and the texture condition determination represents whether the texture class data is all less than the texture class parameter threshold value.

[0070] It should be noted that the final contrast threshold value, the final entropy threshold value and the roughness threshold value of the textile and garment pattern predefined in the preset database are set by professionals.

[0071] The specific expression of the textile and garment pattern texture class defect judgment model is as follows:

[0072] ;

[0073] In the formula, bi represents the judgment result of whether the i-th textile and garment pattern has a structure class defect, Di represents the final contrast of the i-th textile and garment pattern, Ei represents the final entropy of the i-th textile and garment pattern, Fi represents the roughness of the i-th textile and garment pattern, D' represents a final contrast threshold value, E' represents a final entropy threshold value, and F' represents a roughness threshold value. Among them, 1 corresponds to a first judgment value of the texture class defect, and 0 corresponds to a second judgment value of the texture class defect.

[0074] The third type is the color class defect judgment data, and the specific acquisition method is as follows: obtaining a predefined color class parameter threshold value from the database, the color class parameter threshold value including a color difference threshold value and a color histogram difference threshold value; and performing color condition determination on the color class data and the color class parameter threshold value to obtain corresponding color class defect judgment data, thereby effectively identifying quality problems of the textile and garment pattern such as color deviation, color fading, misprinting and uneven dyeing, and the color condition determination represents whether the color class data is all less than the color class parameter threshold value.

[0075] It should be noted that the color difference value threshold and the color histogram difference value threshold of the predefined textile garment pattern in the preset database are set by professionals.

[0076] The specific expression of the textile garment pattern color class defect judgment model is as follows:

[0077]

[0078] In the formula, γi represents the judgment result of whether the i-th textile garment pattern has a structural color class defect, Gi represents the color difference of the i-th textile garment pattern, Hi represents the color histogram difference value of the i-th textile garment pattern, G' represents the color difference value threshold, H' represents the color histogram difference value threshold. Among them, 1 corresponds to the first judgment value of the color class defect, and 0 corresponds to the second judgment value of the color class defect.

[0079] In this embodiment, the defect parameter threshold (structural class parameter threshold, texture class parameter threshold, and color class parameter threshold) judgment mechanism is used to replace the traditional defect judgment method based on artificial experience, which has stronger objectivity and standardization ability; at the same time, the automatic perception ability of pattern details and precision is also enhanced, which is helpful for high-end fabric quality screening; and the system also has the color quality control ability to adapt to different batches and different printing and dyeing processes.

[0080] Further, the specific way of evaluating the defect severity of each textile garment pattern is as follows:

[0081] P1, if each textile garment pattern has a texture class defect, extract the texture class data of each textile garment pattern, combine the corresponding texture class parameter threshold for relative deviation processing, and perform coupling operation based on the correlation between the relative deviation processing result and the texture defect severity coefficient to obtain the texture defect severity coefficient of each textile garment pattern. The texture class parameter threshold includes the final contrast threshold, the final entropy threshold and the roughness threshold.

[0082] The texture defect severity coefficient calculation formula is:

[0083]

[0084] Among them, ηi represents the texture defect severity coefficient of the i-th textile garment pattern, ai represents the final contrast of the i-th textile garment pattern, bi represents the final entropy value of the i-th textile garment pattern, hi represents the roughness of the i-th textile garment pattern, a' represents the final contrast threshold, b' represents the final entropy threshold, h' represents the roughness threshold, and e represents the natural constant.

[0085] ​​In the formula, as the difference between the texture category data and the corresponding texture category parameter threshold value increases, the corresponding texture defect severity coefficient also increases, indicating that the texture category defect of the corresponding textile pattern is more serious.

[0086] P2, if each textile pattern has a color category defect, the color category data of each textile pattern is extracted, a relative deviation operation is performed in combination with the corresponding color category parameter threshold value, a coupling operation is performed according to the correlation between the result of the relative deviation operation and the color defect severity coefficient, and the color defect severity coefficient is obtained. The color category parameter threshold value includes a color difference value threshold and a color histogram difference value threshold.

[0087] The formula for calculating the color defect severity coefficient is:

[0088] ;

[0089] Where μi represents the color defect severity coefficient of the ith textile pattern, fi represents the color difference value of the ith textile pattern, gi represents the color histogram difference value of the ith textile pattern, f' represents the color difference value threshold, and g' represents the color histogram difference value threshold.

[0090] In the formula, as the difference between the color category data and the corresponding color category parameter threshold value increases, the corresponding color defect severity coefficient also increases, indicating that the color category defect of the corresponding textile pattern is more serious.

[0091] P3, based on the texture defect severity coefficient and the color defect severity coefficient of each textile pattern, a summation operation is performed to obtain the defect severity coefficient of each textile pattern.

[0092] In this embodiment, first, the relative deviation analysis of the texture data is beneficial to objectively quantify the severity of different texture defects, and the relative deviation of the color category defect is accurately calculated, which is also helpful for effectively identifying color difference and color deviation. Finally, by comprehensively calculating the severity of the texture category and the color category defect, it is helpful to evaluate the textile pattern quality in multiple dimensions. Through this method, the comprehensive evaluation of the textile pattern is more accurate, and the evaluation result provides a quantitative basis for subsequent automatic classification, product screening and quality control.

[0093] Further, it is judged whether each textile pattern is repaired, and the specific steps are as follows: obtaining the defect severity coefficient of each textile pattern, comparing it with the defect severity coefficient threshold of the textile pattern stored in the preset database, if the defect severity coefficient of a certain textile pattern is lower than the defect severity coefficient threshold of the textile pattern stored in the database, it is determined that the textile pattern is repaired, otherwise it is determined that the textile pattern is not repaired.

[0094] It should be noted that the flaw severity coefficient threshold of the textile garment pattern stored in the preset database is set by a professional.

[0095] In this embodiment, by directly comparing the flaw severity coefficient with the flaw severity coefficient threshold in the preset database, not only the artificial subjective judgment is eliminated, but also the standardization and digitization level of quality management is improved; at the same time, the rework determination is automatically completed without relying on artificial detection, the detection speed and production efficiency are improved, and compared with the traditional determination method based on experience, the rework determination based on data is more objective, which is helpful to avoid misjudgment (unnecessary rework) or missed judgment (defective products flowing into the market) of the textile garment pattern.

[0096] It should be added that after rework repair, it also includes flaw repair allocation and detection, and the specific process is as follows: the repair type of the textile garment pattern after rework repair is judged, if it is single flaw repair, the textile garment pattern is allocated to the corresponding repair assembly line, and the rework frequency is limited based on the preset rework frequency obtained from the preset database to prevent the repeated rework of the scrapped garment pattern, and the repair assembly line includes a texture repair assembly line and a color repair assembly line; if it is double flaw repair, texture type flaw repair detection is performed after texture type flaw repair, and color type flaw repair detection is performed after color type flaw repair, if rework repair is still performed after the first rework repair, and it is double flaw repair, the rework frequency of the corresponding textile garment pattern is limited according to the preset rework frequency limit of the preset proportion, otherwise the rework limit number is mapped according to the current rework frequency.

[0097] It should be explained that the preset rework frequency and the preset proportion are both preset by a professional and stored in the preset database; and in the preset database, the rework frequency is input into the frequency mapping model, and the corresponding rework limit number is output, the frequency mapping model is a pre-trained logistic regression model for fitting the mapping relationship between the rework frequency and the rework limit number, the rework frequency training data is input into the logistic regression model, and the frequency mapping model is trained based on the least square method through the scikit-learn framework, the rework frequency training data includes the rework frequency in the historical period and the rework limit number set by the preset worker according to the rework frequency.

[0098] In this embodiment, according to the flaw type, the corresponding repair assembly line is automatically allocated, the processing efficiency of flaw identification is improved, and resource waste is avoided; at the same time, detection is performed after each repair to ensure that the repair effect meets the standard and prevent defective products from flowing into the market; and an upper limit mechanism of rework frequency is introduced to prevent the same pattern from being repeatedly repaired and the finished product from being scrapped; and this method helps to screen out patterns that are difficult to repair or have poor repair effect after multiple repairs in advance, avoids defective products from flowing into the market, and improves the consistency and product quality control of the terminal product.

[0099] In summary, the embodiment of the present application determines whether there is a structural defect by obtaining the structural data of each textile garment pattern, if there is a structural defect, structural defect feedback correction is performed, in the structural defect detection, the structural problem which seriously affects the quality of the garment can be quickly located, and the structural defect of the textile garment pattern is determined preferentially, thereby realizing rough screening of the textile garment pattern, improving the efficiency of screening the textile garment pattern, otherwise, the texture data and color data of each textile garment pattern are obtained, whether there is a texture defect and a color defect is determined respectively, for the pattern without structural defect, the texture and color defects are further detected in depth, the gradually in-depth detection process improves the detection efficiency, avoids complex detection of all patterns, saves the calculation resources and time cost, and meanwhile ensures the comprehensiveness of detection, if there is a texture defect or a color defect, the defect severity is evaluated based on the texture data and color data of each textile garment pattern to determine whether each textile garment pattern is repaired, otherwise, the textile garment pattern is qualified, thereby realizing comprehensive quantification of the defect severity of each textile garment pattern, improving the scientificity and rationality of decision-making, guiding production more effectively, reducing the cost waste caused by inaccurate defect judgment, and further improving the accuracy of quality judgment of the textile garment pattern, thereby effectively solving the problem of low accuracy of quality judgment of the textile garment pattern caused by inaccurate defect identification in the prior art.

[0100] Those skilled in the art will appreciate that embodiments of the application can be supplied as a method, a system, or a computer program product. Accordingly, the application can be embodied in the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the application can be embodied in the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk memory, CD-ROMs, optical storage media, etc.) having computer usable program code embodied therein.

[0101] The present application is described with reference to the flowcharts and / or block diagrams of the method, device (system), and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the flowcharts and / or block diagrams. Figure 1 The function specified in one flow or multiple flows and / or blocks Figure 1 The device that implements the function specified in one block or multiple blocks.

[0102] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the Figure 1 function specified in the flow or flows and / or blocks Figure 1 of the block or blocks.

[0103] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions that are executed on the computer or other programmable apparatus provide steps for implementing the Figure 1 function specified in the flow or flows and / or blocks Figure 1 of the block or blocks.

[0104] Although preferred embodiments of the application have been described herein, substitutions and alterations are possible in view of the teachings of this application. Accordingly, the appended claims are intended to encompass all such substitutions and alterations. It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application cover the modifications and variations of this application provided they come within the scope of the appended claims and their equivalents.

[0105] It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application cover the modifications and variations of this application provided they come within the scope of the appended claims and their equivalents.

Claims

1. A textile and garment defect pattern data processing method based on a defect judgment model, characterized in that: The following steps are involved: S1, obtaining structural data of each textile and clothing pattern to determine whether each textile and clothing pattern has structural defects. If each textile and clothing pattern has structural defects, feedback and correction of the structural defects are performed; otherwise, S2 is executed; S2, obtaining texture data and color data of each textile and clothing pattern, and determining whether each textile and clothing pattern has texture defects and color defects respectively. If each textile and clothing pattern has texture defects or color defects, then executing S3; otherwise, the textile and clothing pattern is deemed qualified; S3, based on the texture data and color data of each textile and clothing pattern, evaluating the severity of defects of each textile and clothing pattern to determine whether each textile and clothing pattern needs to be reworked and repaired; The specific process of judging whether each textile garment pattern has structural defects is as follows: Extracting structural data of each textile and garment pattern, wherein the structural data includes hole area, drawing length, and warp and weft density; Inputting the structural data into a textile and clothing pattern structural defect judgment model of the textile and clothing pattern, and outputting structural defect judgment data of each textile and clothing pattern, wherein the structural defect judgment data includes a first structural defect judgment value and a second structural defect judgment value; If the textile and clothing pattern structural defect judgment model outputs a first structural defect judgment value, it is determined that the corresponding textile and clothing pattern has structural defects and structural defect feedback correction is performed; if the textile and clothing pattern structural defect judgment model outputs a second structural defect judgment value, it is determined that the corresponding textile and clothing pattern does not have structural defects; The specific method of obtaining the structural defect judgment data is as follows: Obtaining predefined structural parameter thresholds from a database, wherein the structural parameter thresholds include a hole area threshold, a wire drawing length threshold, and a warp and weft density threshold; Performing structural condition determination on the structural data and the structural parameter threshold to obtain corresponding structural defect determination data, wherein the structural condition determination indicates whether the structural data are all not greater than the structural parameter threshold; The structural defect feedback correction mentioned above also includes correction classification, and the specific steps are as follows: C1, based on the structural data, compares the structural classification thresholds stored in a preset database, wherein the structural classification thresholds include a cumulative threshold for hole area, a cumulative threshold for wire drawing length, and a cumulative threshold for warp and weft density; C2: If the hole area of ​​each textile and garment pattern is greater than the cumulative hole area threshold, the corresponding textile and garment pattern is transmitted to the hole scrapping conveyor belt; otherwise, it is transmitted to the hole repairing conveyor belt and C3 is executed. If the number of textile and garment patterns transmitted to the hole scrapping conveyor belt exceeds the preset cumulative number of holes, the preset production personnel are prompted to test the textile and garment fabrics to detect whether there are any quality problems with the fabrics; C3: If the drawing length of each textile and clothing pattern is greater than the drawing length cumulative threshold, the corresponding textile and clothing pattern is transmitted to the drawing scrap conveyor belt; otherwise, it is transmitted to the drawing repair belt and C4 is executed. If the number of textile and clothing patterns transmitted to the drawing scrap conveyor belt exceeds the preset drawing cumulative number, the corresponding textile and clothing pattern is marked as defective; C4: If the warp and weft density of each textile and garment pattern is greater than the warp and weft density cumulative threshold, the corresponding textile and garment pattern is transmitted to the density scrapping conveyor belt; otherwise, it is transmitted to the density repair belt. If the number of textile and garment patterns transmitted to the density scrapping conveyor belt exceeds the preset density cumulative number, the preset production personnel are prompted to check whether the process parameters are correct, and the process parameters include weaving tension and weft feeding speed; The specific steps of determining whether each textile and garment pattern has texture defects and color defects are as follows: P1, extracting texture data of each textile and clothing pattern, wherein the texture data includes final contrast, final entropy value and roughness; P2 is input into the textile and clothing pattern texture defect judgment model, and outputs texture defect judgment data of each textile and clothing pattern, wherein the texture defect judgment data includes a first texture defect judgment value and a second texture defect judgment value; P3: If the textile and clothing pattern texture defect judgment model outputs a first texture defect judgment value, it is determined that the corresponding textile and clothing pattern has texture defects; if the textile and clothing pattern texture defect judgment model outputs a second texture defect judgment value, it is determined that the corresponding textile and clothing pattern does not have texture defects; P4, determine whether there are color defects on each textile and clothing pattern.

2. The method for processing textile and garment defect pattern data based on a defect judgment model according to claim 1, characterized in that: The acquisition of structural data of each textile and clothing pattern also includes determining the flatness of the textile and clothing fabric. The specific steps are as follows: Acquiring fabric smoothness parameters, wherein the fabric smoothness parameters include grayscale gradient, local directional feature entropy value, and edge curvature; Determination is made based on the fabric flatness parameter and the fabric flatness preset data obtained from the preset database, wherein the fabric flatness preset data includes a grayscale gradient preset value, a local directional feature entropy value preset value, and an edge curvature preset value; If there is a fabric flatness parameter lower than the fabric flatness preset data, the preset staff will be prompted to adjust the flatness of the textile and clothing fabric, otherwise the flatness of the textile and clothing fabric will be fed back as qualified.

3. The method for processing textile and garment defect pattern data based on a defect judgment model according to claim 1, characterized in that: The specific process of judging whether each textile and garment pattern has color defects is as follows: Extracting color data of each textile and clothing pattern, wherein the color data includes color difference and color histogram difference value; Inputting color data into a textile and clothing pattern color defect judgment model, and outputting color defect judgment data for each textile and clothing pattern, wherein the color defect judgment data includes a first color defect judgment value and a second color defect judgment value; If the textile and clothing pattern color defect judgment model outputs a first color defect judgment value, it is determined that the corresponding textile and clothing pattern has color defects. If the textile and clothing pattern color defect judgment model outputs a second color defect judgment value, it is determined that the corresponding textile and clothing pattern does not have color defects.

4. The method for processing textile and garment defect pattern data based on a defect judgment model according to claim 1, characterized in that: The specific method of obtaining the texture defect judgment data is as follows: Obtain predefined texture parameter thresholds from a database, wherein the texture parameter thresholds include a final contrast threshold, a final entropy threshold, and a roughness threshold; Performing texture condition determination on the texture data and the texture parameter threshold to obtain corresponding texture defect determination data, wherein the texture condition determination indicates whether the texture data is not greater than the texture parameter threshold; The specific method of obtaining the color defect judgment data is as follows: Obtaining predefined color class parameter thresholds from a database, wherein the color class parameter thresholds include a color difference value threshold and a color histogram difference value threshold; The color data and the color parameter threshold are subjected to color condition determination to obtain corresponding color defect determination data, wherein the color condition determination indicates whether the color data are all not greater than the color parameter threshold.

5. The method for processing textile and garment defect pattern data based on a defect judgment model according to claim 1, characterized in that: The specific method for evaluating the severity of defects of each textile and garment pattern is as follows: If there are texture defects in each textile and clothing pattern, the texture data of each textile and clothing pattern is extracted, and relative deviation processing is performed based on the corresponding texture parameter threshold. A coupling operation is performed based on the correlation between the relative deviation processing result and the texture defect severity coefficient to obtain the texture defect severity coefficient of each textile and clothing pattern; If there are color defects in each textile and clothing pattern, the color data of each textile and clothing pattern is extracted, and a relative deviation operation is performed based on the corresponding color parameter threshold. A coupling operation is performed based on the correlation between the result of the relative deviation operation and the color defect severity coefficient to obtain the color defect severity coefficient; Based on the texture defect severity coefficient and the color defect severity coefficient of each textile and clothing pattern, a summation operation is performed to obtain the defect severity coefficient of each textile and clothing pattern.

6. The method for processing textile and garment defect pattern data based on a defect judgment model according to claim 5, characterized in that: The specific steps of judging whether each textile and garment pattern needs to be reworked and repaired are as follows: The defect severity coefficient of each textile and clothing pattern is obtained and compared with the defect severity coefficient threshold of the textile and clothing pattern stored in a preset database. If the defect severity coefficient of a textile and clothing pattern is lower than the defect severity coefficient threshold of the textile and clothing pattern stored in the database, it is determined that the textile and clothing pattern is to be reworked and repaired; otherwise, it is determined that the textile and clothing pattern is not to be reworked and repaired.

7. The method for processing textile and garment defect pattern data based on a defect judgment model according to claim 1, characterized in that: The rework and repair process also includes defect repair allocation and inspection. The specific process is as follows: Determine the repair type of the textile and garment pattern to be repaired. If it is a single defect repair, assign the textile and garment pattern to the corresponding repair line and set a limit on the number of reworks based on a preset number of reworks obtained from a preset database to prevent repeated rework of the scrapped garment pattern. The repair lines include a texture repair line and a color repair line. If it is a double defect repair, texture defect repair detection is performed after texture defect repair, and color defect repair detection is performed after color defect repair. If repair rework is still performed after the first rework, and it is a double defect repair, the corresponding textile garment pattern rework number is limited according to the preset rework number limit of the preset ratio. Otherwise, the corresponding rework limit number is mapped based on the current rework number.

Images