A visual inspection system for hot stamping defects of a PUR hot stamping roller machine

Abstract

The present invention relates to the technical field of hot stamping defect detection, and specifically to a hot stamping defect visual detection system for a PUR hot stamping roller machine. The present invention performs visual defect detection by applying a certain stretch to the knitted fabric after the hot stamping process and collecting the hot stamping area image after the stretching; it includes extracting the hot stamping pattern through an image processing module, using a defect recognition and analysis module to analyze the integrity, uniformity, position deviation and adhesion of the hot stamping pattern on the knitted fabric, and then evaluating the degree of hot stamping defects with the help of predetermined rules and thresholds. The present invention also analyzes the correlation between defects and working parameters, and adjusts the hot stamping processing parameters of the PUR hot stamping roller machine accordingly; it also summarizes the detection data and analysis results and transmits them to a display device for reference and judgment by professionals. In view of the particularity of knitted fabrics as hot stamping objects, the present invention improves the hot stamping defect detection effect and efficiency of the PUR hot stamping roller machine by defect evaluation from multiple angles.

Description

Technical Field

[0001] The invention relates to the technical field of hot stamping defect detection, in particular to a hot stamping defect visual detection system for a PUR hot stamping roller machine. Background Art

[0002] Compared with traditional hot stamping methods, PUR (moisture-curing reactive polyurethane hot melt adhesive) hot stamping process is used in mass production, precise pattern transfer and decoration processing due to its high efficiency, long-lasting and environmentally friendly hot stamping effect. The PUR hot stamping roller machine, which uses PUR hot melt adhesive as the bonding material, uses a roller to enable the hot stamping operation to be completed at a high speed, thereby achieving large-scale production. Common hot stamping materials include paper products, leather products and textile products. Due to the physical elasticity of paper and leather products, the PUR hot stamping roller machine can achieve a more uniform hot pressing hot stamping effect, but it is different for hot stamping of textiles. Textile fabrics are usually elastic and have uneven surfaces, especially knitted fabrics. Their fiber structure and surface pores make their adsorption and adhesion to hot stamping films far inferior to those of paper and leather products.

[0003] In the prior art, the inspection of hot stamping processing quality is usually carried out on the hot stamping objects with low elasticity and relatively stable physical structure, such as paper products, leather products, glass products and metal products. For the hot stamping objects with changeable physical structure and large surface fluctuations such as knitted fabrics, the defect detection methods of the prior art are difficult to achieve more accurate detection results.

[0004] Based on the above problems, the present application proposes a visual inspection system for hot stamping processing defects of a PUR hot stamping roller machine. Summary of the invention

[0005] The purpose of the present invention is to provide a PUR hot stamping roller machine's hot stamping defect visual detection system, which performs visual defect detection by applying a certain stretch to the knitted fabric after hot stamping and collecting the hot stamping area image after stretching; including extracting the hot stamping pattern through the image processing module, using the defect recognition and analysis module to analyze the integrity, uniformity, position deviation and adhesion of the hot stamping pattern on the knitted fabric, and then using the predetermined rules and thresholds to evaluate the degree of hot stamping defects. The present invention also analyzes the correlation between defects and working parameters, and adjusts the hot stamping processing parameters of the PUR hot stamping roller machine based on this; and also summarizes the detection data and analysis results and transmits them to the display device for reference and judgment by professionals.

[0006] To achieve the above object, the present invention provides the following technical solutions, including: The hot stamping image acquisition module: applies a specified stretch to the knitted fabric output by the hot stamping roller; sets a multi-angle directional light source to collect an image of the hot stamping area of ​​the knitted fabric after stretching, which is recorded as the first image; Image processing module: preprocessing the first image into a second image; the preprocessing steps include denoising and contrast enhancement; using edge detection algorithm to extract the boundary of the hot stamping pattern, and segmenting the hot stamping pattern according to the difference between the boundary and the background area; Defect recognition and analysis module: calculate the ratio of the hot stamping pattern to the hot stamping area and compare it with the standard ratio to obtain the integrity of the hot stamping pattern; calculate the mean difference of the color values ​​of the feature points to obtain the uniformity of the hot stamping pattern; use geometric shape matching to identify the position deviation of the hot stamping pattern; analyze the image gradient of the hot stamping pattern to detect edge changes and obtain the adhesion of the hot stamping pattern; classify and evaluate the detected defects by setting rules and thresholds; Feedback and correction module: conduct correlation analysis between the identified defects and the working parameters of the PUR hot stamping rolling machine; automatically adjust the working parameters according to the defect identification results and correlation analysis; Detection result transmission module: summarizes the defect identification results, correlation analysis results and working parameters, generates a defect analysis reference and transmits it to the display device.

[0007] Furthermore, the preprocessing includes using Gaussian filtering to remove environmental interference and impurities on the surface of the knitted fabric in the first image, and using histogram equalization to highlight the difference between the hot stamping pattern and the background.

[0008] Furthermore, an edge detection algorithm is used to extract an edge image of the hot stamping pattern from the second image, and the brightness value difference between the hot stamping pattern and the background area is identified based on edge information of the edge image, and the hot stamping pattern is segmented based on the brightness value difference.

[0009] Furthermore, the integrity analysis step of the hot stamping pattern includes: for the segmented binary image of the hot stamping pattern, calculating the contour area of ​​the hot stamping pattern through a contour detection algorithm; calculating the ratio of the contour area to the area of ​​the first image; comparing the ratio with the standard ratio to obtain the integrity score of the hot stamping pattern.

[0010] Furthermore, the uniformity analysis step of the hot stamping pattern includes: selecting a number of color feature points in the area of ​​the hot stamping pattern, obtaining the color mean of all the color feature points; calculating the root mean square error based on the color value of the color feature point and the color mean. , get the uniformity score of the hot stamping pattern based on the root mean square error ; ; in, is the maximum tolerable error defined.

[0011] Furthermore, the position deviation analysis step of the hot stamping pattern includes: obtaining a first geometric shape of the standard hot stamping pattern, extracting a second geometric shape of the hot stamping pattern from the binary image of the hot stamping pattern; calculating the coordinate deviation of the contour points of the first geometric shape and the second geometric shape, and obtaining a position deviation score of the hot stamping pattern. ; ; in, is the mean deviation of all contour points, is the maximum permissible deviation defined; For the The coordinate deviation of the contour point is determined by the coordinate position of the contour point in the second geometric shape. and the coordinate position of the contour point in the second geometric shape Calculated; is the number of contour points.

[0012] Furthermore, the step of analyzing the adhesion of the hot stamping pattern includes: using an image gradient algorithm to obtain a gradient map in the second image, defining points whose gradient values ​​exceed a preset gradient threshold as edge points, and Calculate the mean of the gradient difference and obtain the adhesion score of the hot stamping pattern based on the mean of the gradient difference ; ; in, is the mean value of gradient difference, The gradient value of the edge point With preset gradient threshold It is calculated that is the number of edge points; is the maximum gradient value defined and used to normalize the adhesion score.

[0013] Further, the step of classifying and evaluating defects includes: the evaluation includes classification evaluation and comprehensive evaluation; the step of classification evaluation includes: the defect categories include integrity defects, color difference defects, position deviation defects and adhesion defects; when the score of the defect category is lower than the corresponding threshold, the defect category and the corresponding score are directly output; The steps of comprehensive evaluation include: when the scores of all defect categories are not lower than the corresponding thresholds, all scores are combined into the hot stamping processing score of the batch of knitted fabrics.

[0014] Furthermore, the steps of correlation analysis include: continuously obtaining the working parameters and defect categories of the PUR hot stamping rolling machine, and using regression analysis to model the correlation between the defect category and the working parameters; the correlation model is: ; in, For the The regression analysis model of defect categories is For the Working parameters; is a constant term, For the The regression coefficients corresponding to the working parameters; is the error term, is the number of working parameters; Get real-time defect category , according to the regression analysis model The relevant working parameters of the PUR hot stamping roller machine are automatically corrected through the relevant working parameters; Correction of working parameters The calculation formula is: ; in, For the The regression coefficients of the working parameters are for The change in rating.

[0015] Furthermore, defect information is extracted from the defect identification and analysis module, including defect category and corresponding score, defect location coordinates, and hot stamping processing score; working parameters, defect category and cause, correlation coefficient and correction content are extracted from the feedback and correction module, and graphical reference data is generated and transmitted to the display device.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention applies a specified stretch to the knitted fabric and collects the images of the gold-stamping area after stretching at multiple angles, which helps to reduce the image acquisition error caused by the undulation of the fabric surface. The light source setting ensures the clarity and stability of the image acquisition image; the knitted fabric after the specified stretch is collected to ensure the stability of the gold-stamping and perform gold-stamping processing defect detection. The present invention extracts the gold-stamping pattern on the knitted fabric from the background through a visual processing method, which reduces the processing time and improves the accuracy of subsequent defect recognition and analysis.

[0017] 2. The present invention solves the limitations of existing detection methods when dealing with knitted fabrics and other textile objects that need to be stamped by performing multi-dimensional analysis on the hot stamping patterns on knitted fabrics. The multi-dimensional analysis includes evaluating the integrity through the proportion of the hot stamping pattern area to help find the problem of missing hot stamping; evaluating the uniformity through the color mean difference to help find the color difference problem; identifying the position deviation through the comparison of geometric shape contour points to ensure the correct positioning of the hot stamping pattern; and evaluating the adhesion effect of the hot stamping pattern on the knitted fabric through image gradient analysis. Combining the multi-dimensional detection results through classification evaluation and comprehensive evaluation can judge the degree of defects from single to comprehensive, thereby improving the accuracy and reliability of defect detection.

[0018] 3. The present invention provides a closed-loop control solution for defect detection of PUR hot stamping roller machine through feedback and correction module. First, through correlation analysis, a correlation relationship is established between the generation of defects and the working parameters of the hot stamping roller machine, and regression analysis modeling is used to enable the correlation relationship to be applied to subsequent production; then, when defects are identified, the working parameters of the PUR hot stamping roller machine are automatically adjusted according to the root cause of the defect to optimize the hot stamping process, reduce the probability of repeated defects, and improve the consistency and stability of the hot stamping processing quality of the PUR hot stamping roller machine. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a flow chart of a visual inspection system for hot stamping defects of a PUR hot stamping roller machine; Figure 2 It is a structural schematic diagram of a visual inspection system for hot stamping processing defects of a PUR hot stamping roller machine; Figure 3 Schematic diagram of defect recognition and analysis method based on visual inspection technology. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] In view of the problem that the existing technology is difficult to achieve a relatively accurate detection effect for knitted fabrics, which have a changeable physical structure and large surface fluctuations, this application proposes a PUR hot stamping machine hot stamping defect visual detection system. The specific process diagram is shown in Figure 1 The following will describe the specific implementation of the system proposed in this application through Example 1 and Example 2.

[0022] Embodiment 1 Reference Figure 2 , Figure 2 The present invention is a structural schematic diagram of a visual inspection system for hot stamping processing defects of a PUR hot stamping roller machine. The system includes a hot stamping image acquisition module, an image processing module, a defect recognition and analysis module, a feedback and correction module, and a detection result transmission module.

[0023] Furthermore, the hot stamping image acquisition module is used to apply a specified stretch to the knitted fabric output by the hot stamping roller; set a light source, and collect images of the hot stamping area of ​​the knitted fabric after stretching from multiple angles, which are recorded as the first image. Specifically: Due to the elastic properties of knitted fabrics, in addition to the pattern deformation or damage caused by the hot stamping process, there are also subsequent deformation or damage problems caused by poor hot stamping adhesion. For example, the integrity and uniformity of the pattern after hot stamping are very good, but it will fall off and crack after a slight stretch, which shows that the adhesion of hot stamping is very poor. This embodiment chooses to perform image acquisition after simple stretching of the knitted fabric after hot stamping for subsequent defect detection in order to ensure that the hot stamping adhesion of the knitted fabric tested meets the basic requirements.

[0024] This embodiment chooses to apply a certain stretch to the knitted fabric after stamping in the same direction as the rotation direction of the stamping roller equipment. The stretching range is set between 10% and 15% of the length of the knitted fabric in the stretching direction to ensure that the stamping pattern will not affect the subsequent detection effect due to excessive or insufficient stretching.

[0025] Set up directional light sources at multiple angles to collect images of the gold stamping area of ​​the knitted fabric after stretching, which are recorded as the first image. When using directional light sources at multiple angles, the purpose is to avoid shadows and reflections when collecting images and ensure the clarity of the collected images.

[0026] Furthermore, the image processing module is used to pre-process the first image into a second image; an edge detection algorithm is used to extract the boundary of the hot stamping pattern, and the hot stamping pattern is segmented according to the difference between the boundary and the background area. Specifically: The preprocessing step in this embodiment processes the hot stamping fabric image by calling relevant tools in OpenCV, including first using a 3×3 Gaussian kernel to denoise the first image, and then using CLAHE (adaptive histogram equalization) to enhance the contrast of the denoised first image to obtain a second image; then, using the Canny edge detection algorithm to extract the boundary of the hot stamping pattern of the second image, and then binarizing the boundary image to separate the hot stamping pattern from the image background.

[0027] The threshold of Canny edge detection can be set according to actual needs. In this embodiment, the low threshold and the high threshold are set to 50 and 150 respectively.

[0028] This embodiment removes random noise in the image by denoising the collected first image to ensure the accuracy of subsequent processing; the purpose of contrast enhancement is to enhance the brightness difference between the hot stamping pattern and the background in the image, making the pattern more prominent and facilitating subsequent edge detection and segmentation.

[0029] This embodiment calls the Canny edge detection in OpenCV to highlight the edge of the hot stamping pattern in the second image, and then separates the edge of the hot stamping pattern from the background based on the edge information, extracts a clear and clean hot stamping pattern, and provides clear and accurate image data for subsequent defect detection.

[0030] Furthermore, the defect recognition and analysis module is used to obtain the integrity, uniformity, position deviation and adhesion of the hot stamping pattern from the hot stamping pattern image segmented from the second image; and to classify and evaluate the detected defects by setting rules and thresholds. Specifically: Reference Figure 3 , Figure 3 Schematic diagram of defect recognition and analysis method based on visual inspection technology. For the segmented binary image of the hot stamping pattern, this embodiment calls the function in OpenCV to extract the contour information of the hot stamping pattern and calculates the contour area of ​​the hot stamping pattern, which is recorded as ; The area of ​​the first image is obtained by multiplying the width and height of the first image, denoted as .

[0031] By calculating the contour area With the first image area Ratio The ratio of the area of ​​the standard hot stamping pattern to the area of ​​the first image is recorded as the standard ratio ; The ratio Ratio to standard Compare and get the integrity score of the hot stamping pattern. The calculation formula is: ; In the calculation formula of the integrity score, when the ratio Ratio to standard If the deviation is too large, the integrity score will decrease, indicating that the integrity of the hot stamping pattern is low.

[0032] This embodiment verifies the integrity by comparing the area with the largest proportion of the hot stamping pattern, and can accurately evaluate the degree of deviation between the hot stamping pattern and the standard pattern; by calculating the ratio of the hot stamping pattern contour area to the total image area and comparing it with the standard ratio, it can quantify the coverage of the pattern, help detect whether the hot stamping area is complete, and whether there are real and blank areas, thereby reducing errors in human judgment and improving the accuracy of integrity detection.

[0033] A number of color feature points are selected in the area of ​​the hot stamping pattern. The selected feature points are usually located at representative positions of the hot stamping pattern and cover different areas of the hot stamping pattern to ensure that the overall color characteristics of the pattern can be represented. In this embodiment, the selection of feature points includes but is not limited to the edge and middle area of ​​the hot stamping pattern, highlight and shadow areas, and areas of different shapes and textures.

[0034] At the selected color feature point, obtain the RGB color value of each point, and then calculate the color mean of all color feature points; calculate the root mean square error based on the color value of the color feature point and the color mean , the root mean square error calculation process is expressed as: ; in, is the number of selected color feature points, For the The RGB color value of the color feature point, is the color mean of all color feature points.

[0035] Obtaining the uniformity score of the hot stamping pattern based on the root mean square error : ; in, is the maximum tolerable error defined, and The larger the difference, the better the uniformity of the hot stamping pattern, and vice versa.

[0036] This embodiment obtains the color distribution of the entire area of ​​the hot stamping pattern by selecting color feature points and calculating the color mean of all color feature points; and then evaluates the uniformity of the hot stamping pattern by calculating the root mean square error of all color feature points, which can achieve more accurate quality evaluation. This method can systematically measure the color consistency of the hot stamping area and help staff to grasp the quality of the hot stamping pattern in real time.

[0037] Obtain edge contour points of a standard hot stamping pattern, recorded as a first geometric shape, and extract edge contour points of the hot stamping pattern from the binary image of the hot stamping pattern, recorded as a second geometric shape; Obtain the corresponding contour points of the first geometric shape and the second geometric shape, and calculate the coordinate deviation between each group of corresponding contour points to obtain the position deviation score of the hot stamping pattern ; ; in, is the mean deviation of all contour points, is the maximum permissible deviation defined; For the The coordinate deviation of the contour point is determined by the coordinate position of the contour point in the second geometric shape. The coordinate position of the contour point in the second geometric shape Calculated; is the number of contour points.

[0038] In-position bias scoring It can be seen from the formula that The value range is [0, 1]; when the deviation mean of all contour points When smaller, Close to 1, indicating that the position deviation of the hot stamping pattern is small; When larger, If it is close to 0, it means that the pattern position deviation is large.

[0039] This embodiment accurately evaluates the stamping position accuracy of the pattern by calculating the geometric shape deviation of the stamping pattern. By comparing the coordinate position deviation of the edge contour points, the deformation of the geometric shape can be accurately measured. The size of the position deviation reflects the degree of deformation of the stamping pattern on the knitted fabric. This improves the processing defect detection accuracy of knitted fabrics as stamping objects.

[0040] The gradient value of each pixel in the second image is calculated to obtain a gradient map. The gradient value reflects the brightness change of each pixel in the image. If the gradient value exceeds the preset gradient threshold , which means that the point is the edge point of the hot stamping pattern. The edge points of the hot stamping pattern are extracted from the gradient image.

[0041] By edge points and preset gradient threshold Calculate the mean gradient difference of all edge points, and obtain the adhesion score of the hot stamping pattern based on the mean gradient difference ; ; in, is the mean of the gradient differences, given by The gradient value of the edge point With preset gradient threshold It is calculated that is the number of edge points; is the maximum gradient value defined and used to normalize the adhesion score.

[0042] According to the adhesion scoring formula, the greater the gradient difference of the edge points, the greater the adhesion score value, indicating that the pattern has better adhesion; the smaller the gradient difference of the edge points, the smaller the adhesion score value, indicating that the hot stamping pattern has worse adhesion.

[0043] This embodiment can accurately identify the adhesion of the hot stamping pattern on the knitted fabric by analyzing the gradient change of the hot stamping pattern in the second image. The greater the difference between the edge gradient of the hot stamping pattern and the preset threshold, the more obvious the edge of the pattern is and the clearer the contrast with the background is; these performances mean that the hot stamping pattern has better adhesion on the knitted fabric. This method can improve the accuracy of hot stamping adhesion detection on knitted fabrics.

[0044] The evaluation is carried out according to the above-mentioned completeness score, uniformity score, position deviation score and adhesion score, including classification evaluation and comprehensive evaluation.

[0045] The steps of classification evaluation include: defect categories include integrity defects, corresponding to integrity scores; color difference defects, corresponding to uniformity scores; position deviation defects, corresponding to position deviation scores; adhesion defects, corresponding to adhesion scores; Set a threshold for each defect category score. When the score of a defect category is lower than the corresponding threshold, directly output the defect category and the corresponding score. The steps of comprehensive evaluation include: when the scores of all defect categories are not lower than the corresponding threshold, all scores are combined into the hot stamping score of the batch of knitted fabrics; the process of obtaining the hot stamping score is expressed as: ; in, Score the hot stamping process. , , and The weights of the integrity score, uniformity score, position deviation score and adhesion score are set and adjusted by those skilled in the art according to actual conditions.

[0046] A factory used a PUR hot stamping roller machine numbered A001 to perform hot stamping on a batch of knitted fabrics. This embodiment uses the hot stamping defect detection method to perform hot stamping defect detection. The batch of knitted fabrics was divided into three groups for testing. The defect detection results of the batch of knitted fabrics obtained are given in Table 1. In this embodiment, , , and Set to 0.3, 0.2, 0.2 and 0.3 respectively.

[0047] Table 1 Detection results of defects in hot stamping of knitted fabrics

[0048] According to the data in Table 1, the integrity of the hot stamping pattern of this batch of knitted fabrics is good; the uniformity fluctuates, but the overall situation is good; the fluctuation of the position deviation score is small, and the overall score is high, indicating that the position deviation of the hot stamping pattern of this batch of knitted fabrics is small and the deformation is good; the adhesion score is high, indicating that the hot stamping adhesion of this batch of knitted fabrics is strong. In summary, the hot stamping processing quality of this batch of knitted fabrics is good overall, reflecting the consistency and stability of the overall hot stamping processing.

[0049] This embodiment achieves accurate evaluation of the defects of hot stamping patterns on knitted fabrics by combining classification evaluation and comprehensive evaluation. Classification evaluation can give full attention to key issues according to the different types of defects identified; comprehensive evaluation weights and integrates each score when the scores of all defect types are within their corresponding reasonable ranges to form an overall hot stamping processing score, which improves the accuracy and flexibility of hot stamping processing defect detection and provides a reliable basis and support for optimization and decision-making in actual production.

[0050] Embodiment 2 This embodiment describes a specific implementation method of a feedback and correction module of a hot stamping defect visual inspection system for a PUR hot stamping roller machine.

[0051] Furthermore, the feedback and correction module is used to perform correlation analysis between the identified defects and the working parameters of the PUR hot stamping rolling machine; according to the defect identification results and the correlation analysis results, the working parameters of the PUR hot stamping rolling machine are automatically adjusted. Specifically: The working parameters and defect categories of the PUR hot stamping roller machine are continuously obtained, and the correlation between the defect category and the working parameters is modeled using regression analysis; the correlation model is: ; in, For the The scores of defect categories, For the Working parameters; is a constant term, For the The regression coefficients corresponding to the working parameters; is the error term, is the number of working parameters. Based on the analysis of the continuously collected data, the regression coefficient is calculated through correlation modeling, which represents the degree of influence of the working parameters on the defect category.

[0052] Get real-time defect category , obtained based on the regression analysis model Corresponding relevant working parameters, and automatically correct the current working parameters of the PUR hot stamping roller machine through the relevant working parameters; No. Correction of working parameters The calculation formula is: ; in, For the The regression coefficients of the working parameters are for The change in rating.

[0053] In order to verify the effectiveness of the method proposed in this application, this embodiment is compared with the existing hot stamping defect visual detection method. The comparison method is described as follows: Method 1: Based on traditional image processing technology, defects are identified by comparing image similarities; it is commonly used for conventional ironing materials with flat structures and smooth surfaces, such as paper, plastic cards, etc. Method 2: Use machine learning algorithms to extract the features of hot stamping patterns and train models to automatically identify and classify hot stamping defects. This method is commonly used for hot stamping materials that are smooth and have stable structures, such as paper, leather, and glass. Method 3: Use deep learning methods to extract image features for training and optimization to identify fine defects; suitable for high-hardness ironing materials such as paper and leather, as well as high-density textiles; The comparison results are shown in Table 2.

[0054] Table 2 Comparative test results of hot stamping defect detection methods

[0055] As can be seen from Table 2, the method proposed in this application can achieve good detection results in the task of detecting defects in hot stamping of knitted fabrics. Compared with the other three methods, the method proposed in this application takes into account the particularity of knitted fabrics, which are hot stamping objects with high elasticity and large surface pores. Through visual detection methods and image processing methods, the adhesion ability, hot stamping integrity, uniformity and deformation resistance of hot stamping are analyzed and calculated in an all-round way. Therefore, it can show good results in the task of detecting defects in hot stamping of knitted fabrics. In addition, the method proposed in this application also avoids a large amount of data training, making the method easy to implement.

[0056] This embodiment continuously collects the working parameters and defect categories of the PUR hot stamping roller machine, builds a model through a regression analysis model, analyzes the influence of each working parameter on each defect category, and obtains the correlation between the working parameters and the defect category. According to the correlation, the cause of the real-time defect is quickly identified and the relevant parameters are adjusted. This method can improve the follow-up capability of hot stamping processing defect detection and realize closed-loop feedback after the processing defect is detected.

[0057] Furthermore, the detection result transmission module is used to summarize the defect identification results, correlation analysis results and working parameters, generate a defect analysis reference and transmit it to the display device. Specifically: Defect information is extracted from the defect identification and analysis module, including defect category and corresponding score, defect location coordinates, and hot stamping processing score; real-time defect category, related working parameters and correction amount are extracted from the feedback and correction module, and graphical reference data is generated and transmitted to the display device.

[0058] This embodiment realizes the coordination of defect monitoring, analysis and correction by automatically extracting defect detection information and generating graphical reference data. The graphical reference data can help the staff to grasp the working condition and work quality of the PUR hot stamping roller machine in real time during the hot stamping process, and adjust the working parameters in time to ensure the stability of the hot stamping quality of knitted fabrics.

[0059] The present application proposes a visual detection system for defects in hot stamping of knitted fabrics, which is low in accuracy and needs to be improved in the hot stamping process of the PUR hot stamping roller machine. Knitted fabrics have high elasticity, large surface pores, large surface undulations and roughness, while the prior art usually shows good detection effects on hot stamping objects with low elasticity and stable physical structure, and is difficult to be well applied to the defect detection of hot stamping of knitted fabrics. The present application proposes to perform defect detection of hot stamping on hot stamping knitted fabrics after slight stretching, and adopts a variety of image processing technologies to comprehensively evaluate the hot stamping quality of the PUR hot stamping roller machine from the integrity, uniformity, deformation deviation and adhesion performance of the hot stamping pattern. In addition, the present application also proposes to analyze the correlation between the identified defect categories and the working parameters of the PUR hot stamping roller machine, find out the causes of the defects and establish a regression analysis model; then analyze the causes of real-time defects through the regression analysis model, so as to adjust and optimize the hot stamping process in a timely manner; finally, the defect analysis results and working parameters are summarized and transmitted to the display device for the staff to intuitively understand the production situation. Compared with the existing technology, the method proposed in this application has a more accurate defect effect in the detection of hot stamping defects in knitted fabrics, can also analyze the causes of defects, and does not require a large amount of training data, thereby improving the accuracy and efficiency of visual detection of hot stamping defects of PUR hot stamping roller machines.

[0060] Embodiment 3 This embodiment provides additional explanations on the defect categories in the hot stamping process defect detection of the PUR hot stamping roller machine, the corresponding hot stamping roller machine working parameters, and the cause analysis.

[0061] When integrity defects occur, possible causes include: Insufficient hot stamping temperature results in uneven thermal bonding and the hot stamping pattern falls off; Insufficient hot stamping pressure and unsatisfactory heat transfer effect, which results in the hot stamping pattern not being completely fixed on knitted fabrics with large fluctuations; The hot stamping speed is too fast, and the heat transfer and bonding action are insufficient.

[0062] When uniformity defects occur, possible causes include: The hot stamping temperature is uneven, and the temperature of some areas is greatly different from that of other parts, which will lead to poor hot stamping. Uneven pressure, the hot stamping pressure is unevenly distributed on the fabric, and the heat transfer is incomplete in some areas; The hot stamping time is too short and the heat transfer effect is insufficient, resulting in the pattern being lighter or unclear in some areas.

[0063] When position deviation defects occur, possible causes include: Equipment positioning deviation: The positioning system of the hot stamping equipment is not calibrated correctly; Uneven hot stamping pressure: affects the directionality of the hot stamping pattern; Fabric stretch: The fixing effect of knitted fabric is not good, resulting in deviation of hot stamping pattern.

[0064] When adhesion defects occur, possible causes include: The hot stamping temperature is too low / too high: If the temperature is too low, the pattern will not adhere firmly, and if the temperature is too high, the pattern will be overly adhered and cannot be transferred smoothly; The hot stamping time is too short: the thermal transfer effect is not sufficient.

[0065] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A visual inspection system for hot stamping defects of a PUR hot stamping roller machine, characterized in that: include: Hot stamping image acquisition module: applies a specified stretch to the knitted fabric output by the hot stamping roller; Setting a multi-angle directional light source to collect an image of the hot stamping area of ​​the knitted fabric after stretching, which is recorded as the first image; Image processing module: preprocessing the first image into a second image; the preprocessing steps include denoising and contrast enhancement; using an edge detection algorithm to extract the boundary of the hot stamping pattern, and segmenting the hot stamping pattern according to the difference between the boundary and the background area; Defect recognition and analysis module: calculates the ratio of the hot stamping pattern to the hot stamping area and compares it with the standard ratio to obtain the integrity of the hot stamping pattern; calculates the mean difference of the color values ​​of the feature points to obtain the uniformity of the hot stamping pattern; Using geometric shape matching to identify the position deviation of the hot stamping pattern; analyzing the image gradient of the hot stamping pattern to detect edge changes and obtain the adhesion of the hot stamping pattern; classifying and evaluating the detected defects by setting rules and thresholds; Feedback and correction module: conduct correlation analysis between the identified defects and the working parameters of the PUR hot stamping rolling machine; Automatically adjusting the working parameters according to the defect identification results and the correlation analysis; Detection result transmission module: summarizes the defect identification result, the correlation analysis result and the working parameter, generates a defect analysis reference and transmits it to the display device.

2. According to claim 1, a PUR hot stamping roller machine hot stamping defect visual inspection system is characterized in that: The preprocessing includes using Gaussian filtering to remove environmental interference and impurities on the surface of the knitted fabric in the first image, and using histogram equalization to obtain the difference characteristics of the hot stamping pattern from the background area.

3. The visual inspection system for hot stamping defects of a PUR hot stamping roller machine according to claim 1 is characterized in that: An edge detection algorithm is used to extract an edge image of the hot stamping pattern from the second image, a brightness difference between the hot stamping pattern and the background area is identified based on edge information of the edge image, and the hot stamping pattern is segmented based on the brightness difference.

4. The visual inspection system for hot stamping defects of a PUR hot stamping roller machine according to claim 1 is characterized in that: The integrity analysis steps of the hot stamping pattern include: For the segmented binary image of the hot stamping pattern, the contour area of ​​the hot stamping pattern is calculated by a contour detection algorithm; the ratio of the contour area to the area of ​​the first image is calculated; the ratio is compared with a standard ratio to obtain an integrity score of the hot stamping pattern.

5. The visual inspection system for hot stamping defects of a PUR hot stamping roller machine according to claim 1 is characterized in that: The uniformity analysis step of the hot stamping pattern comprises: Select a number of color feature points in the area of ​​the hot stamping pattern to obtain the color mean of all the color feature points; calculate the root mean square error based on the color value of the color feature point and the color mean , obtain the uniformity score of the hot stamping pattern according to the root mean square error ; ； in, is the maximum tolerable error defined.

6. The visual inspection system for hot stamping defects of a PUR hot stamping roller machine according to claim 1 is characterized in that: The position deviation analysis step of the hot stamping pattern comprises: Obtain a first geometric shape of a standard hot stamping pattern, extract a second geometric shape of the hot stamping pattern from a binary image of the hot stamping pattern; calculate the coordinate deviation of the contour points of the first geometric shape and the second geometric shape, and obtain a position deviation score of the hot stamping pattern ; ； in, is the mean deviation of all contour points, is the maximum permissible deviation defined; For the The coordinate deviation of the contour point is determined by the coordinate position of the contour point in the second geometric shape. and the coordinate position of the contour point in the second geometric shape Calculated; is the number of contour points.

7. The visual inspection system for hot stamping defects of a PUR hot stamping roller machine according to claim 1 is characterized in that: The adhesion analysis step of the hot stamping pattern comprises: The image gradient algorithm is used to obtain the gradient map in the second image, and the points whose gradient values ​​exceed the preset gradient threshold are defined as edge points. Calculate the mean value of the gradient difference, and obtain the adhesion score of the hot stamping pattern according to the mean value of the gradient difference ; ； in, is the mean value of the gradient difference, given by The gradient value of the edge point With the preset gradient threshold It is calculated that is the number of edge points; is the maximum gradient value defined and used to normalize the adhesion score.

8. The visual inspection system for hot stamping defects of a PUR hot stamping roller machine according to claim 1 is characterized in that: The steps to classify and evaluate the defects include: The evaluation includes classification evaluation and comprehensive evaluation; the classification evaluation step includes: the defect categories include integrity defects, uniformity defects, position deviation defects and adhesion defects; when the score of the defect category is lower than the corresponding threshold, the defect category and the corresponding score are directly output; The comprehensive evaluation step includes: when the scores of all the defect categories are not lower than the corresponding threshold value, all the scores are combined into a hot stamping score of the knitted fabric.

9. The visual inspection system for hot stamping defects of a PUR hot stamping roller machine according to claim 1, characterized in that: The steps of the correlation analysis include: The working parameters and defect categories of the PUR hot stamping roller machine are continuously obtained, and the correlation between the defect category and the working parameters is modeled using regression analysis; the regression analysis model is: ； in, For the The score of each defect category, For the The working parameters; is a constant term, For the The regression coefficient corresponding to each of the working parameters; is the error term, is the number of the working parameters; Get real-time defect category , according to the regression analysis model, Corresponding relevant working parameters, and automatically correcting the current working parameters of the PUR hot stamping roller machine through the relevant working parameters; No. The correction amount of the working parameters The calculation formula is: ； in, For the The regression coefficients of the working parameters are for The change in rating.

10. The visual inspection system for hot stamping defects of a PUR hot stamping roller machine according to claim 1, characterized in that: The generation steps of the defect analysis reference include: Defect information is extracted from the defect recognition and analysis module, including defect category and corresponding score, defect location coordinates, and hot stamping processing score; real-time defect category, relevant working parameters and correction amount are extracted from the feedback and correction module, and graphical reference data is generated and transmitted to the display device.

Images