Surface quality visual inspection system for packaging paper presswork

Abstract

The invention discloses a surface quality visual detection system for a packaging paper printed matter, and relates to the technical field of printed matter detection. When the quality tracing signaling is triggered, the fault matching model is constructed according to the defect indexes of the current K groups of unqualified printed matters, meanwhile, the historical data is extracted from the database to construct the fault historical model, and the comprehensive similarity index is obtained by calculating the area similarity value and the shape similarity value of the fault matching model and the fault historical model. According to the method, the fault historical model with the highest comprehensive similarity index is obtained, the fault reasons are extracted, and the K groups of fault reasons are matched, counted and sorted to form the fault reason set to serve as the pre-estimated defect reasons of the printing equipment, so that the potential fault reasons of the printing equipment can be quickly positioned in combination with historical data, and compared with manual troubleshooting of equipment parts, the efficiency is improved. And the tracing efficiency is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of printed matter detection, and specifically to a visual inspection system for the surface quality of wrapping paper printed matter. Background Art

[0002] In the production process of wrapping paper printing, the surface quality of printed matter is crucial. It not only affects the aesthetics and overall image of the product, but may also affect consumers' trust in the product, and thus is related to the market competitiveness of the product.

[0003] However, the current visual inspection systems for the surface quality of wrapping paper printed matter still have the following deficiencies in actual application:

[0004] In actual production, the existing visual inspection systems only judge whether the product is qualified, and can no longer meet the enterprise's demand for refined management of product quality. They cannot accurately quantify and evaluate various defects of wrapping paper printed matter, such as scratches, patterns, and inks, which is not conducive to the enterprise's in-depth analysis of the root causes of product quality problems.

[0005] When faced with a large number of unqualified products, the existing detection methods are difficult to quickly and accurately trace the cause of the printing equipment failure in combination with historical data. When there are batch printing defects, the enterprise can only manually check each component of the equipment, which is time-consuming, laborious, and inefficient, and cannot solve the problems in production in a timely manner.

[0006] Therefore, a surface quality detection system for wrapping paper printed matter is introduced. Summary of the Invention

[0007] The purpose of the present invention is to solve the problems pointed out in the background art, and to propose a visual inspection system for the surface quality of wrapping paper printed matter.

[0008] The purpose of the present invention can be achieved by the following technical solutions: A visual inspection system for the surface quality of wrapping paper printed matter, comprising:

[0009] Defect recognition module: Analyze the image data collected from the wrapping paper printed matter, and determine various defect indices of the image data of the wrapping paper printed matter according to the analysis results; among which various defect indices include physical defect index M1, pattern defect index M2, and ink defect index M3; based on various defect indices of the image data of the wrapping paper printed matter, determine the printing quality index Ger of the image data of the wrapping paper printed matter.

[0010] Defect handling module: Set the quality threshold index corresponding to the printing quality index Ger, and classify each group of packaging printed products into qualified printed products and unqualified printed products according to the set quality threshold index. At the same time, continuously monitor the classification results of each group of packaging paper printed products. If the classification results of K groups of packaging paper printed products reach unqualified printed products during continuous monitoring, trigger a quality traceability signal and send it to the defect traceability module; where K > 5;

[0011] Defect traceability module: When triggering the quality traceability signal, extract the printing quality index Ger of K groups of unqualified printed products, and execute corresponding steps to determine the estimated defect cause of the printing equipment.

[0012] As a preferred embodiment of the present invention, determine the physical defect index M1 of the packaging paper printed product image data, specifically:

[0013] Identify scratch defect features, wrinkle defect features, and hole defect features from the image data of the packaging paper printed product;

[0014] Divide the image data of the packaging paper printed product into each image sub-region, count the total number of pixels of scratch defect features, the total number of pixels of wrinkle defect features, and the total number of pixels of hole defect features in each image sub-region, and convert them into scratch feature area, wrinkle feature area, and hole feature area based on the resolution of the image;

[0015] Set the weight coefficients corresponding to the scratch feature area, wrinkle feature area, and hole feature area respectively, multiply the scratch feature area, wrinkle feature area, and hole feature area of each image sub-region by the corresponding set weight coefficients, and then sum to obtain the defect sub-value of each image sub-region;

[0016] Set the weight coefficients corresponding to different graphic sub-regions, extract the defect sub-values of each image sub-region, multiply them by the corresponding weight coefficients respectively, and then sum to obtain the defect comprehensive value of the packaging paper printed product;

[0017] Preset the interval where the comprehensive value corresponding to the defect comprehensive value is located. Each interval where the comprehensive value is located corresponds to a physical defect score of the packaging paper printed product; Match the defect comprehensive value of the packaging paper printed product with the corresponding interval where the comprehensive value is located to obtain the physical defect score of the packaging paper printed product;

[0018] And use the obtained physical defect score as the physical defect index M1 of the packaging paper printed product image data.

[0019] As a preferred embodiment of the present invention, determine the pattern defect index M2 of the packaging paper printed product image data, specifically:

[0020] Extract the qualified image data corresponding to the wrapper printed matter image data, align the wrapper printed matter image data with the qualified image data, perform feature point detection and descriptor calculation on the two sets of image data respectively to obtain a feature point set and the corresponding descriptors; perform feature point matching on the two sets of image data to obtain a set of matching points; select M matching point pairs from the set of matching points to obtain a transformation matrix from the wrapper printed matter image data to the qualified image data;

[0021] Use the obtained transformation matrix to transform the wrapper printed matter image so that the wrapper printed matter is aligned with the qualified image data in terms of spatial position, compare the two sets of aligned image data pixel by pixel, and calculate the mean square error between the wrapper printed matter image data and the qualified image data;

[0022] Set a reference threshold corresponding to the mean square error. If the mean square error between the wrapper printed matter image data and the qualified image data is less than the reference threshold, take a preset value as the pattern defect index M2 of the wrapper printed matter image data; if the mean square error between the wrapper printed matter image data and the qualified image data is greater than the reference threshold, calculate the difference between the mean square error and the reference threshold, denoted as the deviation degree value;

[0023] Preset the interval where the degree value corresponding to the deviation degree value is located. Each interval where the degree value is located corresponds to a pattern defect score of the wrapper printed matter; match the deviation degree value of the wrapper printed matter with the corresponding interval where the degree value is located to obtain the pattern defect score of the wrapper printed matter; and take the obtained pattern defect score as the pattern defect index M2 of the wrapper printed matter image data.

[0024] As a preferred embodiment of the present invention, determine the ink defect index M3 of the wrapper printed matter image data, specifically:

[0025] Identify the ink dot defect features and ink spot defect features from the image data of the wrapper printed matter;

[0026] Count the total number of pixels of the ink dot defect features and the total number of pixels of the ink spot defect features in the wrapper printed matter image data, and convert them into the ink dot feature area and the ink spot feature area based on the resolution of the image;

[0027] Set the weight coefficients corresponding to the ink dot feature area and the ink spot feature area respectively, multiply the ink dot feature area and the ink spot feature area of the wrapper printed matter image data by the corresponding set weight coefficients respectively, and then sum them to obtain the ink defect value of each image sub-region, denoted as Q1;

[0028] Set the standard color of the wrapping paper print, calculate the average color value of the wrapping paper print in the Lab color space, denoted as L1, a1, and b1; extract the average color value of the standard color in the Lab color space, denoted as L2, a2, and b2;

[0029] According to the formula Calculate the color difference evaluation value Q2 of the wrapping paper print;

[0030] Substitute the ink defect value Q1 and the color difference evaluation value Q2 of the wrapping paper print into the formula Perform weighted calculation to obtain the ink performance value Q3 of the wrapping paper print; Q1 允许 and Q2 允许 respectively represent the maximum allowable value of ink defects and the maximum allowable value of color difference evaluation; where α1 and α2 are the influence weight factors of the ink defect value Q1 and the color difference evaluation value Q2 respectively;

[0031] Preset the interval where the performance value corresponding to the ink performance value Q3 is located. Each group of performance value intervals corresponds to an ink defect score of the wrapping paper print; match the ink performance value Q3 of the wrapping paper print with the corresponding performance value interval to obtain the ink defect score of the wrapping paper print;

[0032] And take the obtained ink defect score as the ink defect index M3 of the wrapping paper print image data.

[0033] As a preferred embodiment of the present invention, determine the printing quality index Ger of the wrapping paper print image data, specifically:

[0034] Extract the physical defect index M1, the pattern defect index M2, and the ink defect index M3 of the wrapping paper print; set the weight coefficients corresponding to the physical defect index M1, the pattern defect index M2, and the ink defect index M3 respectively. Multiply the physical defect index M1, the pattern defect index M2, and the ink defect index M3 of the wrapping paper print by the corresponding set weight coefficients respectively, and then sum to obtain the printing quality index Ger of the wrapping paper print image data.

[0035] As a preferred embodiment of the present invention, classify each group of packaging prints into qualified prints and unqualified prints according to the set quality threshold index, specifically:

[0036] Analyze the printing quality index Ger of each group of wrapping paper prints and compare it with the set quality threshold index. If the printing quality index Ger of a certain group of wrapping paper prints is higher than the set quality threshold index, it is classified as an unqualified print, otherwise it is classified as a qualified print.

[0037] As a preferred embodiment of the present invention, when triggering the quality traceability signaling, the printing quality index Ger of K groups of unqualified printed products is extracted, and the corresponding steps are executed, specifically as follows:

[0038] S1: A fault case database corresponding to the printing equipment is pre-constructed. The physical defect index, pattern defect index, and ink defect index of the packaging printed products produced when each historical fault case of the printing equipment occurred are obtained. The fault causes that led to each historical fault of the printing equipment are obtained, and the corresponding fault causes are combined with the corresponding physical defect index, pattern defect index, and ink defect index and stored in the fault case database.

[0039] S2: From the printing quality index Ger of K groups of unqualified printed products, the physical defect index M1, pattern defect index M2, and ink defect index M3 of the K groups of unqualified printed products are obtained. Three rays are drawn starting from the origin as the center point; the lengths of the three rays respectively correspond to the values of the physical defect index M1, pattern defect index M2, and ink defect index M3. Using the endpoints of the three rays as marking points, the three groups of marking points are sequentially connected to form a closed polygon, and the formed polygon is used as the fault matching model of the unqualified printed products.

[0040] As a preferred embodiment of the present invention, when triggering the quality traceability signaling, the printing quality index Ger of K groups of unqualified printed products is extracted, and the corresponding steps are executed, further including:

[0041] S3: Similarly to step S2, the physical defect index, pattern defect index, and ink defect index of the packaging printed products produced when each historical fault case in the fault case database occurred are extracted, and a polygon is constructed, and the constructed polygon is used as the fault historical model corresponding to each historical fault case.

[0042] S4: The areas of the fault matching model and each group of fault historical models are obtained. Taking the area of the fault matching model as the numerator and the area of the fault historical model as the denominator, the ratio between the two areas is calculated to obtain the area similarity value F1 between the fault matching model and each group of fault historical models.

[0043] By calculating the Euclidean distance, the closest distance values of the three sets of ray coordinates between the fault matching model and each group of fault historical models are obtained, and the maximum value of the three closest distance values is selected as the shape similarity value F2 between the fault matching model and each group of fault historical models.

[0044] The area similarity value F1 and shape similarity value F2 of the fault matching model and each group of fault historical models are normalized and then substituted into the formula Perform calculations to obtain the comprehensive similarity index F3 between the fault matching model and each group of fault history models; where η1 and η2 are the influence weight factors of the area similarity value F1 and the shape similarity value F2 respectively.

[0045] As a preferred embodiment of the present invention, determining the estimated defect cause of the printing equipment specifically includes:

[0046] S5: Select the fault history model with the highest comprehensive similarity index F3, and extract the fault cause corresponding to the fault history model; match the K groups of fault causes with each other. If the same fault cause appears, count the number of occurrences of the same fault cause, and sort the fault causes from high to low according to the number of occurrences. After sorting, integrate them into a fault cause set, and use the fault cause set as the estimated defect cause of the printing equipment.

[0047] Compared with the prior art, the beneficial effects of the present invention are:

[0048] The present invention processes the image data of the wrapping paper printed matter, identifies various defect features such as scratches, wrinkles, holes, ink dots, and ink spots, counts the total number of relevant pixels and converts it into an area, matches the defect scores by presetting different intervals, and finally determines the physical defect index, pattern defect index, ink defect index, and printing quality index. These quantified indexes can accurately reflect the defect degree of the wrapping paper printed matter, meet the enterprise's demand for refined management of product quality, and help the enterprise deeply analyze the root cause of product quality problems;

[0049] When the quality traceability signaling is triggered, the present invention constructs a fault matching model according to the defect indexes of the current K groups of unqualified printed matters, and at the same time extracts historical data from the database to construct a fault history model. By calculating the area similarity value and shape similarity value between the fault matching model and the fault history model, the comprehensive similarity index is obtained, the fault history model with the highest comprehensive similarity index is obtained, its fault cause is extracted, and the K groups of fault causes are matched, counted, and sorted to form a fault cause set, which is used as the estimated defect cause of the printing equipment. This method combines historical data and can quickly locate the potential fault causes of the printing equipment, greatly improving the traceability efficiency compared with manual inspection of equipment components. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] For the convenience of those skilled in the art to understand, the present invention will be further described below with reference to the accompanying drawings.

[0051] Figure 1 It is the principle block diagram of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0052] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts belong to the scope of protection of the present invention.

[0053] Please refer to Figure 1 As shown, a surface quality visual inspection system for packaging paper printed matter includes a defect recognition module, a defect processing module, and a defect traceability module;

[0054] The defect recognition module is used to analyze the image data collected from the packaging paper printed matter, and determine various defect indexes of the image data of the packaging paper printed matter according to the analysis results; among them, various defect indexes include a physical defect index M1, a pattern defect index M2, and an ink defect index M3; based on various defect indexes of the image data of the packaging paper printed matter, determine the printing quality index Ger of the image data of the packaging paper printed matter.

[0055] It should be noted that the image data is obtained by installing a high-resolution industrial camera above the transmission path of the packaging paper printed matter, which can meet the clear capture of the surface details of the packaging paper printed matter, and preprocess the collected image data before analysis. The Gaussian filtering algorithm is used to denoise the image, and the size of the Gaussian kernel is adjusted according to the noise situation of the image. Then, the color image is converted into a grayscale image. Next, the histogram equalization algorithm is used to enhance the grayscale image. By redistributing the gray values of the image, the gray histogram of the image becomes more uniform, and the contrast between the defects and the background in the image is enhanced.

[0056] To determine the physical defect index M1 of the image data of the packaging paper printed matter, specifically:

[0057] Identify scratch defect features, wrinkle defect features, and hole defect features from the image data of the packaging paper printed matter;

[0058] Based on the visual attention of the current packaging paper printed matter, divide the image data of the packaging paper printed matter into each image sub-region, count the total number of pixels of scratch defect features, wrinkle defect features, and hole defect features in each image sub-region, and convert them into scratch feature area, wrinkle feature area, and hole feature area based on the resolution of the image.

[0059] Set the weight coefficients corresponding to the scratch feature area, wrinkle feature area, and hole feature area respectively, multiply the scratch feature area, wrinkle feature area, and hole feature area of each image sub-region by the corresponding set weight coefficients, and then sum to obtain the defect sub-values of each image sub-region;

[0060] According to the visual attention of different image sub-regions, set the weight coefficients corresponding to different image sub-regions, extract the defect sub-values of each image sub-region, multiply them by the corresponding weight coefficients respectively, and then sum to obtain the comprehensive defect value of the wrapping paper print;

[0061] Preset the interval where the comprehensive value corresponding to the comprehensive defect value is located. Each interval where the comprehensive value is located corresponds to a physical defect score of the wrapping paper print; the physical defect score range is set from 1 to 10 and is a positive integer. The higher the comprehensive defect value, the higher the physical defect score obtained by matching; match the comprehensive defect value of the wrapping paper print with the corresponding interval where the comprehensive value is located to obtain the physical defect score of the wrapping paper print;

[0062] And use the obtained physical defect score as the physical defect index M1 of the wrapping paper print image data;

[0063] It should be noted that, for example:

[0064] The process of scratch recognition is as follows:

[0065] Adopt the Canny edge detection algorithm. First, perform Gaussian smoothing on the preprocessed image data to reduce the influence of noise; then calculate the gradient magnitude and direction of the image; then perform non-maximum suppression to eliminate the spurious responses brought by edge detection; finally, through double-threshold processing and edge connection, determine the true edges. Scratches usually appear as slender edges, and the position of the scratches can be initially located through the detected edge information;

[0066] Perform morphological operations on the image after edge detection, such as dilation and erosion. The dilation operation can expand the edges of the scratches to connect the broken scratches; the erosion operation can remove some small interfering edges. By reasonably selecting the size and shape of the structural elements for dilation and erosion, the complete shape of the scratches can be better extracted;

[0067] Extract the features of the scratches, such as length, width, area, etc., set appropriate thresholds, and judge whether the extracted edges are scratches; for example, if the length of the edge exceeds a certain threshold and the width is relatively narrow, it is considered a scratch. At the same time, the accuracy of scratch detection can be further improved by combining information such as the position and direction of the scratches in the image;

[0068] The process of wrinkle recognition is as follows:

[0069] Wrinkles can cause changes in the texture of the wrapping paper surface. Therefore, texture analysis methods can be used to detect wrinkles. Commonly used texture analysis methods include gray-level co-occurrence matrix and local binary pattern (LBP). The gray-level co-occurrence matrix can describe the spatial distribution relationship of gray levels in an image. By calculating its statistical features, such as contrast, correlation, energy, etc., to determine whether there is a wrinkled area in the image. LBP is a simple and effective texture description operator. By comparing the gray value of the central pixel with that of the neighboring pixels, a binary pattern is generated to reflect the local texture information of the image. Region segmentation, based on the results of texture analysis, uses methods such as threshold segmentation and clustering analysis to segment the image into regions, separating the wrinkled area from the background. Threshold segmentation sets a threshold according to the gray value or texture features of the image to divide the image into different regions. Clustering analysis classifies the pixels in the image according to their features, and pixels belonging to the same class are considered to be in the same region. Feature verification extracts the features of the wrinkled area, such as area, shape, texture complexity, etc. By comparing with a preset wrinkled feature model, it is verified whether the segmented area is a real wrinkle. For example, if the area of the region is large and the texture complexity is high, it is determined to be a wrinkled area;

[0070] The process of void recognition is as follows:

[0071] The preprocessed image is binarized to convert the image into a black-and-white image, highlighting the features of the voids. The commonly used binarization method is the global threshold method, such as the Otsu algorithm, which can automatically calculate the optimal threshold to divide the image into foreground and background. Connected component analysis is performed on the binarized image to find all connected components in the image. Each connected component represents a possible void. By calculating the features of the connected components, such as area, perimeter, centroid, etc., the regions that meet the void features are screened out. For example, an area threshold is set to remove the connected components with too small area, considering them as noise or other interference factors;

[0072] Determine the pattern defect index M2 of the wrapping paper printed image data, specifically:

[0073] Extract the qualified image data corresponding to the wrapping paper printed matter image data, and align the wrapping paper printed matter image data with the qualified image data; the qualified image data is stored in the database, and each wrapping paper printed matter image has a unique identifier corresponding to it (such as product number, batch number, etc.); retrieve the corresponding qualified image data from the database through this identifier; use the feature point extraction algorithm to perform feature point detection and descriptor calculation on the two sets of image data respectively to obtain the feature point set and the corresponding descriptor; use the feature point matching algorithm to perform feature point matching on the two sets of image data to obtain the matching point set; select M matching point pairs from the matching point set; where M is initially set by the technician; use the Random Sample Consensus algorithm to obtain the transformation matrix from the wrapping paper printed matter image data to the qualified image data;

[0074] Use the obtained transformation matrix to transform the wrapping paper printed matter image so that the wrapping paper printed matter is aligned with the qualified image data in terms of spatial position, and compare the two sets of aligned image data pixel by pixel to calculate the mean square error between the wrapping paper printed matter image data and the qualified image data;

[0075] Set the reference threshold corresponding to the mean square error. If the mean square error between the wrapping paper printed matter image data and the qualified image data is less than the reference threshold, take the preset value as the pattern defect index M2 of the wrapping paper printed matter image data; when it is less than the reference threshold, it means that the pattern positions of the two sets of image data completely coincide, and the preset value can be taken as 0 or 1; if the mean square error between the wrapping paper printed matter image data and the qualified image data is greater than the reference threshold, calculate the difference between the mean square error and the reference threshold, and record it as the deviation degree value;

[0076] Preset the interval where the degree value corresponding to the deviation degree value is located. Each interval where the degree value is located corresponds to a pattern defect score of a wrapping paper printed matter; the pattern defect score range is set from 1 to 10 and is a positive integer. The higher the deviation degree value, the higher the pattern defect score obtained by matching; match the deviation degree value of the wrapping paper printed matter with the corresponding interval where the degree value is located to obtain the pattern defect score of the wrapping paper printed matter;

[0077] And take the obtained pattern defect score as the pattern defect index M2 of the wrapping paper printed matter image data;

[0078] Determine the ink defect index M3 of the wrapping paper printed matter image data, specifically:

[0079] Identify the ink dot defect features and ink spot defect features from the image data of the wrapping paper printed matter;

[0080] Statistically calculate the total number of pixels of the ink dot defect features and the total number of pixels of the ink spot defect features in the wrapping paper printed matter image data, and convert them into the ink dot feature area and the ink spot feature area based on the resolution of the image;

[0081] Set the weight coefficients corresponding to the characteristic area of ink dots and the characteristic area of ink spots respectively. Multiply the characteristic area of ink dots and the characteristic area of ink spots in the image data of the wrapping paper print by the corresponding set weight coefficients, and then sum them to obtain the ink defect value of each image sub-region, denoted as Q1;

[0082] Set the standard color of the wrapping paper print according to the standard color card, calculate the average color values of the wrapping paper print in the Lab color space, denoted as L1, a1, and b1; extract the average color values of the standard color in the Lab color space, denoted as L2, a2, and b2;

[0083] According to the formula Calculate the color difference evaluation value Q2 of the wrapping paper print;

[0084] Substitute the ink defect value Q1 and the color difference evaluation value Q2 of the wrapping paper print into the formula Perform weighted calculation to obtain the ink performance value Q3 of the wrapping paper print; Q1 允许 、Q2 允许 respectively represent the maximum allowable value of ink defects and the maximum allowable value of color difference evaluation; they are set according to the printing quality requirements; where α1 and α2 are the influence weight factors of the ink defect value Q1 and the color difference evaluation value Q2 respectively;

[0085] Preset the interval where the performance value corresponding to the ink performance value Q3 is located. Each interval where the performance value is located corresponds to an ink defect score of the wrapping paper print; the range of the ink defect score is set from 1 to 10 and is a positive integer. The higher the ink performance value Q3, the higher the matching ink defect score; match the ink performance value Q3 of the wrapping paper print with the corresponding interval where the performance value is located to obtain the ink defect score of the wrapping paper print;

[0086] And use the obtained ink defect score as the ink defect index M3 of the image data of the wrapping paper print;

[0087] It should be noted that, for example:

[0088] Ink dot defect recognition process:

[0089] Perform binaryzation processing on the preprocessed grayscale image to convert the image into a black-and-white image, highlighting the difference between the ink dots and the background. The Otsu algorithm can be used to automatically determine the optimal threshold to divide the image into two parts: foreground (ink dots) and background;

[0090] Using morphological dilation and erosion operations, the binarized image is processed. First, a dilation operation is performed using a circular or square structuring element to expand the edges of the ink dots and connect the possibly broken parts. Then, an erosion operation is carried out to remove some isolated noise points and make the shape of the ink dots clearer and more accurate.

[0091] Through connected component analysis, each connected black area is regarded as an ink dot.

[0092] Ink spot defect identification process:

[0093] Using methods based on region growing or clustering analysis, the image is segmented to separate the ink spot area from the background. Based on color features and spatial neighborhood relationships, pixels with similar colors and adjacent to each other are merged into one area, thereby identifying the ink spot area.

[0094] Extract the physical defect index M1, pattern defect index M2, and ink defect index M3 of the wrapping paper print. Set the corresponding weight coefficients for the physical defect index M1, pattern defect index M2, and ink defect index M3. Multiply the physical defect index M1, pattern defect index M2, and ink defect index M3 of the wrapping paper print by the corresponding set weight coefficients respectively, and then sum them to obtain the printing quality index Ger of the wrapping paper print image data.

[0095] The defect processing module is used to set the quality threshold index corresponding to the printing quality index Ger, classify each group of wrapping paper prints into qualified prints and unqualified prints according to the set quality threshold index, and continuously monitor the classification results of each group of wrapping paper prints. If the classification results of K groups of wrapping paper prints reach unqualified prints during the continuous monitoring process, a quality traceability signaling is triggered and sent to the defect traceability module; where K > 5, and the specific value is set by technical personnel.

[0096] Specifically:

[0097] Analyze the printing quality index Ger of each group of wrapping paper prints and compare it with the set quality threshold index. If the printing quality index Ger of a certain group of wrapping paper prints is higher than the set quality threshold index, it is classified as an unqualified print; otherwise, it is classified as a qualified print.

[0098] It should be noted that if it is classified as unqualified printed matter, the printing quality index Ger corresponding to the unqualified printed matter is extracted, and the difference is calculated with the quality threshold index. The difference is recorded as the estimated difference. If the estimated difference is lower than the preset reference difference, the image data of the unqualified printed matter is extracted, and the printing quality index Ger is analyzed again. The re-analyzed printing quality index Ger is compared with the quality threshold index. If the re-analyzed printing quality index Ger is lower than the quality threshold index, the average value of the two sets of printing quality index Ger is taken as the final evaluation value, and the result of comparing the final evaluation value with the quality threshold index for classification is used as the final classification result of the corresponding unqualified printed matter.

[0099] The defect traceability module is used to extract the printing quality index Ger of K groups of unqualified printed matter when the quality traceability signaling is triggered, and execute corresponding steps to determine the estimated defect cause of the printing equipment;

[0100] Specifically:

[0101] S1: A failure case database corresponding to the printing equipment is pre-constructed. The physical defect index, pattern defect index, and ink defect index of the packaging printed matter produced when each historical failure case of the printing equipment occurs are obtained. The failure causes that lead to each historical failure of the printing equipment are obtained, and the corresponding failure causes are combined with the corresponding physical defect index, pattern defect index, and ink defect index and stored in the failure case database;

[0102] S2: From the printing quality index Ger of K groups of unqualified printed matter, the physical defect index M1, pattern defect index M2, and ink defect index M3 of K groups of unqualified printed matter are obtained. Three rays are drawn starting from the origin as the center point; the first ray is along the positive x-axis direction and its length corresponds to the value of the physical defect index M1, the second ray forms an angle of 120 degrees with the positive x-axis direction and its length corresponds to the value of the pattern defect index M2, and the third ray forms an angle of 240 degrees with the positive x-axis direction and its length corresponds to the value of the ink defect index M3; the lengths of the three rays respectively correspond to the values of the physical defect index M1, pattern defect index M2, and ink defect index M3. Taking the endpoints of the three rays as marking points, the three groups of marking points are sequentially connected to form a closed polygon, and the formed polygon is used as the failure matching model of the unqualified printed matter;

[0103] It should be noted that assume that from the printing quality index Ger of K groups of unqualified printed matter, a physical defect index of 8, a pattern defect index of 6, and an ink defect index of 5 are obtained;

[0104] Determine the endpoint coordinates of the three rays:

[0105] The first ray: Along the positive x-axis direction, with a length of 8, and its endpoint coordinates are (8,0);

[0106] The second ray: The angle with the positive direction of the x-axis is 120 degrees, and the length is 6; according to the formula for converting polar coordinates to rectangular coordinates, the end coordinates of the second ray can be calculated as (-3, 5.2):

[0107] The third ray: The angle with the positive direction of the x-axis is 240 degrees, and the length is 5; similarly, according to the above formula, the end coordinates of the third ray can be calculated as (-2.5, -4.3).

[0108] Connect these three coordinates in sequence to form a closed polygon, and this polygon is the fault matching model constructed based on this set of defective printing defect indices.

[0109] S3: Similarly to step S2, extract the physical defect indices, pattern defect indices, and ink defect indices of the packaging printed products produced during each historical fault case in the fault case database, and construct a polygon. Take the constructed polygon as the fault historical model corresponding to each historical fault case;

[0110] S4: Obtain the areas of the fault matching model and each group of fault historical models. Use the area of the fault matching model as the numerator and the area of the fault historical model as the denominator to calculate the ratio between the two areas to obtain the area similarity value F1 between the fault matching model and each group of fault historical models;

[0111] Through the calculation of the Euclidean distance, obtain the closest distance values of the three sets of ray coordinates between the fault matching model and each group of fault historical models, and select the maximum value of the three closest distance values as the shape similarity value F2 between the fault matching model and each group of fault historical models;

[0112] It should be noted that the lower the closest distance value, the higher the degree of proximity in the spatial position between the corresponding two groups of models.

[0113] Normalize the area similarity value F1 and the shape similarity value F2 of the fault matching model and each group of fault historical models and substitute them into the formula for calculation to obtain the comprehensive similarity index F3 between the fault matching model and each group of fault historical models; where η1 and η2 are the influence weight factors of the area similarity value F1 and the shape similarity value F2 respectively;

[0114] S5: Select the fault historical model with the highest comprehensive similarity index F3, and extract the fault causes corresponding to the fault historical model; match the K groups of fault causes with each other. If the same fault cause appears, count the number of occurrences of the same fault cause, and sort the fault causes from high to low according to the number of occurrences. After sorting, integrate them into a fault cause set, and use the fault cause set as the estimated defect cause of the printing equipment;

[0115] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments only. Obviously, according to the content of this specification, many modifications and variations can be made. These embodiments are selected and specifically described in this specification in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A visual inspection system for the surface quality of printed packaging paper, characterized in that: include: Defect recognition module: analyzes the image data collected from the packaging paper prints, and determines various defect indexes of the packaging paper prints image data based on the analysis results; The various defect indexes include physical defect index M1, pattern defect index M2 and ink defect index M3; Determine the printing quality index Ger of the packaging paper printed product image data based on various defect indexes of the packaging paper printed product image data; Defect processing module: Set the quality threshold index corresponding to the printing quality index Ger, and classify each group of packaging printed products into qualified printed products and unqualified printed products according to the set quality threshold index, and continuously monitor the classification results of each group of packaging paper printed products. If the classification results of the cumulative number of K groups of packaging paper printed products are unqualified printed products during the continuous monitoring process, the quality traceability signaling is triggered and sent to the defect traceability module; where K>5; Defect tracing module: When the quality tracing signal is triggered, the printing quality index Ger of K groups of unqualified printed products is extracted, and corresponding steps are executed to determine the estimated defect cause of the printing equipment.

2. The surface quality visual inspection system for packaging paper printed products according to claim 1 is characterized in that: Determine the physical defect index M1 of the packaging paper printed product image data, specifically: Identify scratch defect features, wrinkle defect features, and void defect features from image data of packaging paper prints; Divide the image data of the packaging paper print into various image sub-regions, count the total number of pixels of the scratch defect feature, the total number of pixels of the wrinkle defect feature, and the total number of pixels of the cavity defect feature in each image sub-region, and convert them into scratch feature area, wrinkle feature area, and cavity feature area based on the resolution of the image; The weight coefficients corresponding to the scratch feature area, wrinkle feature area and hole feature area are set respectively, and the scratch feature area, wrinkle feature area and hole feature area of ​​each image sub-area are multiplied by the corresponding set weight coefficients respectively, and then the sum is obtained to obtain the defect sub-value of each image sub-area; The weight coefficients corresponding to different graphic sub-regions are set, the defect sub-values ​​of each image sub-region are extracted, and the defect sub-values ​​are multiplied by the corresponding weight coefficients respectively, and then the sum is obtained to obtain the comprehensive defect value of the packaging paper printing product; The interval of the comprehensive value corresponding to the preset defect comprehensive value, each group of comprehensive value intervals respectively corresponds to a physical defect score of the packaging paper printed product; the defect comprehensive value of the packaging paper printed product is matched with the corresponding comprehensive value interval to obtain the physical defect score of the packaging paper printed product; The obtained physical defect score is used as the physical defect index M1 of the packaging paper print image data.

3. The surface quality visual inspection system for packaging paper printed products according to claim 2 is characterized in that: Determine the pattern defect index M2 of the packaging paper printed product image data, specifically: Extract qualified image data corresponding to the packaging paper printed product image data, align the packaging paper printed product image data with the qualified image data, perform feature point detection and descriptor calculation on the two sets of image data respectively, and obtain a feature point set and corresponding descriptors; perform feature point matching on the two sets of image data to obtain a matching point set; Selecting M matching point pairs from the matching point set to obtain a transformation matrix from the packaging paper printed matter image data to the qualified image data; The obtained transformation matrix is ​​used to transform the packaging paper printed product image so that the packaging paper printed product is aligned with the qualified image data in spatial position, and the two aligned sets of image data are compared pixel by pixel to calculate the mean square error between the packaging paper printed product image data and the qualified image data; A reference threshold corresponding to the mean square error is set. If the mean square error between the packaging paper printed product image data and the qualified image data is less than the reference threshold, a preset value is taken as the pattern defect index M2 of the packaging paper printed product image data; if the mean square error between the packaging paper printed product image data and the qualified image data is greater than the reference threshold, the difference between the mean square error and the reference threshold is calculated and recorded as the deviation degree value; The preset deviation degree value corresponds to an interval of degree values, and each group of degree value intervals corresponds to a pattern defect score of a packaging paper print; the deviation degree value of the packaging paper print is matched with the corresponding degree value interval to obtain the pattern defect score of the packaging paper print; and the obtained pattern defect score is used as the pattern defect index M2 of the packaging paper print image data.

4. The surface quality visual inspection system for packaging paper printed products according to claim 3 is characterized in that: Determine the ink defect index M3 of the packaging paper printed product image data, specifically: Identifying ink dot defect features and ink spot defect features from image data of packaging paper prints; Counting the total number of pixels of ink dot defect features and the total number of pixels of ink spot defect features in the packaging paper printed product image data, and converting them into ink dot feature areas and ink spot feature areas based on the resolution of the image; The weight coefficients corresponding to the ink dot characteristic area and the ink spot characteristic area are set respectively, and the ink dot characteristic area and the ink spot characteristic area of ​​the packaging paper printed product image data are multiplied by the corresponding set weight coefficients respectively, and then the sum is obtained to obtain the ink defect value of each image sub-area, which is recorded as Q1; Set the standard color of the printed wrapping paper, calculate the average color value of the printed wrapping paper in the Lab color space, record it as L1, a1 and b1; extract the average color value of the standard color in the Lab color space, record it as L2, a2 ​​and b2; According to the formula Calculate the color difference evaluation value Q2 of the packaging paper print; Substitute the ink defect value Q1 and color difference evaluation value Q2 of the packaging paper print into the formula Perform weighted calculation to obtain the ink performance value Q3 of the packaging paper printing; Q1 允许 、Q2 允许 They represent the maximum allowable value of ink defects and the maximum allowable value of color difference evaluation respectively; α1 and α2 are the influencing weight factors of ink defect value Q1 and color difference evaluation value Q2 respectively; The performance value interval corresponding to the preset ink performance value Q3, each group of performance value intervals respectively corresponds to an ink defect score of the packaging paper printed product; the ink performance value Q3 of the packaging paper printed product is matched with the corresponding performance value interval to obtain the ink defect score of the packaging paper printed product; The obtained ink defect score is used as the ink defect index M3 of the packaging paper print image data.

5. The surface quality visual inspection system for packaging paper printed products according to claim 4, characterized in that: Determine the printing quality index Ger of the packaging paper printed product image data, specifically: Extract the physical defect index M1, pattern defect index M2 and ink defect index M3 of the packaging paper print; set the weight coefficients corresponding to the physical defect index M1, pattern defect index M2 and ink defect index M3, respectively, multiply the physical defect index M1, pattern defect index M2 and ink defect index M3 of the packaging paper print by the corresponding set weight coefficients, and then sum them up to obtain the printing quality index Ger of the packaging paper print image data.

6. The surface quality visual inspection system for packaging paper printed products according to claim 5, characterized in that: Each group of packaging printed products is classified into qualified printed products and unqualified printed products according to the set quality threshold index, specifically: The printing quality index Ger of each group of packaging paper printed products is analyzed and compared with the set quality threshold index. If the printing quality index Ger of a group of packaging paper printed products is higher than the set quality threshold index, it is classified as unqualified printed products, otherwise it is classified as qualified printed products.

7. A surface quality visual inspection system for packaging paper printed products according to claim 6, characterized in that: When the quality traceability signaling is triggered, the printing quality index Ger of the K groups of unqualified printed products is extracted, and the corresponding steps are executed, specifically: S1: pre-build a fault case database corresponding to the printing equipment, obtain the physical defect index, pattern defect index and ink defect index of the packaging printed products produced when each historical fault case corresponding to the printing equipment occurs, obtain the fault cause that causes each historical fault of the printing equipment, and combine the corresponding fault cause with the corresponding physical defect index, pattern defect index and ink defect index and store them in the fault case database; S2: From the printing quality index Ger of K groups of unqualified printed products, obtain the physical defect index M1, pattern defect index M2 and ink defect index M3 of the K groups of unqualified printed products, and draw three rays with the origin as the center point; the lengths of the three rays correspond to the values ​​of the physical defect index M1, the pattern defect index M2 and the ink defect index M3 respectively, and the end points of the three rays are used as marking points, and the three groups of marking points are connected in sequence to form a closed polygon, and the formed polygon is used as the fault matching model of the unqualified printed products.

8. The surface quality visual inspection system for packaging paper printed products according to claim 7, characterized in that: When the quality traceability signal is triggered, the printing quality index Ger of the K groups of unqualified printed products is extracted and corresponding steps are executed, including: S3: Similarly, step S2 extracts the physical defect index, pattern defect index, and ink defect index of the packaging printed products produced when each historical failure case in the failure case database occurs, and constructs a polygon, and uses the constructed polygon as the failure history model corresponding to each historical failure case; S4: Obtain the areas of the fault matching model and each group of fault history models, use the area of ​​the fault matching model as the numerator and the area of ​​the fault history model as the denominator, calculate the ratio between the two groups of areas, and obtain the area similarity value F1 between the fault matching model and each group of fault history models; By calculating the Euclidean distance, the closest distance values ​​of the three groups of ray coordinates between the fault matching model and each group of fault history models are obtained, and the maximum value of the three groups of closest distance values ​​is selected as the shape similarity value F2 between the fault matching model and each group of fault history models; After normalizing the area similarity value F1 and shape similarity value F2 of the fault matching model and each group of fault history models, they are inserted into the formula Calculation is performed to obtain the comprehensive similarity index F3 between the fault matching model and each group of fault history models; wherein η1 and η2 are the influence weight factors of the area similarity value F1 and the shape similarity value F2, respectively.

9. A surface quality visual inspection system for packaging paper printed products according to claim 8, characterized in that: Determine the estimated defect causes of printing equipment, specifically: S5: Select the fault history model with the highest comprehensive similarity index F3, and extract the fault cause of the corresponding fault history model; match K groups of fault causes with each other, if the same fault cause appears, count the number of occurrences of the same fault cause, and sort the fault causes from high to low according to the number of occurrences, and after sorting, integrate them into a fault cause set, and use the fault cause set as the estimated defect cause of the printing equipment.

Images