Method and system for detecting and inhibiting wrinkles in gravure printing

Abstract

The invention provides an intaglio printing wrinkling detection and inhibition method and system, and belongs to the technical field of printing, and the method comprises the steps: carrying out the splicing, preprocessing and feature extraction of linear array data, obtaining feature image data, inputting the feature image data to a pre-trained wrinkling evaluation model, and obtaining a wrinkling region and a pixel-level mask; analyzing the wrinkling direction and the wrinkling density of the wrinkling area according to the pixel-level mask, and dividing the wrinkling area to obtain a plurality of sub-areas; calculating light properties and tension of all the sub-regions to obtain a light property anomaly index and a tension instability index; determining a sub-evaluation coefficient of each sub-region based on the light anomaly index and the tension instability index, and obtaining a comprehensive influence level of the wrinkling region based on the sub-evaluation coefficients; determining a wrinkling inhibition strategy based on the comprehensive influence level and the plurality of sub-regions; and an inhibition instruction is generated based on the wrinkling inhibition strategy, and the printing machine unit corresponding to the wrinkling area is adjusted. According to the invention, the automation of intaglio printing is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of printing, in particular to a method and system for detecting and suppressing wrinkles in intaglio printing. BACKGROUND

[0002] Intaglio printing is widely used in the packaging and publishing industries due to its high print quality and high durability. However, during the printing process, the printed paper is easily affected by factors such as tension fluctuations and mechanical transmission deviations, which can cause paper wrinkles. This not only reduces the pass rate of printed products, but also leads to production accidents such as paper jams and roller wear. Existing wrinkle detection relies on manual visual inspection, which has the disadvantages of low detection efficiency, subjectivity, and inability to intervene in real time. Automatic detection schemes based on single feature recognition cannot achieve pixel-level positioning of the wrinkle area, and the suppression strategy lacks targetedness and can cause secondary defects due to blind adjustment of tension parameters.

[0003] Against this background, there is an urgent need to develop a method and system for detecting and suppressing wrinkles in intaglio printing. SUMMARY

[0004] To solve the above technical problems, the present application provides a method and system for detecting and suppressing wrinkles in intaglio printing.

[0005] The first aspect of the embodiments of the present application provides a method for detecting and suppressing wrinkles in intaglio printing, comprising: acquiring line array data of the printed paper by a line array camera and splicing to obtain image data; preprocessing and feature extraction are performed on the image data to obtain feature image data; inputting the feature image data into a pre-trained wrinkle evaluation model to obtain a wrinkle area and a pixel-level mask thereof; analyzing the wrinkle direction and wrinkle density of the wrinkle area according to the pixel-level mask, and dividing the wrinkle area to obtain a plurality of sub-areas; calculating the optical property and tension of all sub-areas to obtain an optical anomaly index and a tension instability index; determining a sub-evaluation coefficient of each sub-area based on the optical anomaly index and the tension instability index; determining a comprehensive influence level of the wrinkle area according to the distribution characteristics of all sub-evaluation coefficients; determining a suppression strategy for wrinkles based on the comprehensive influence level and the plurality of sub-areas; generating a suppression instruction based on the suppression strategy for wrinkles, and adjusting the corresponding printing machine unit of the wrinkle area based on the suppression instruction.

[0006] In a second aspect, the embodiment of the present application provides a gravure printing wrinkle detection and suppression system, comprising: An image acquisition and processing module is configured to acquire line array data of the printed paper by the line array camera and perform splicing to obtain image data, and perform pre-processing and feature extraction on the image data to obtain feature image data; A wrinkle region determination module is configured to input the feature image data into a pre-trained wrinkle evaluation model to obtain a wrinkle region and a pixel-level mask thereof; A wrinkle region division module is configured to analyze a wrinkle direction and a wrinkle density of the wrinkle region according to the pixel-level mask, and divide the wrinkle region to obtain a plurality of sub-regions; An influence level determination module is configured to calculate optical properties and tension of all sub-regions to obtain an optical abnormality index and a tension instability index, determine a sub-evaluation coefficient of each sub-region based on the optical abnormality index and the tension instability index, and determine a comprehensive influence level of the wrinkle region according to a distribution feature of all sub-evaluation coefficients; A suppression strategy determination module is configured to determine a wrinkle suppression strategy based on the comprehensive influence level and the plurality of sub-regions; An instruction generation and execution module is configured to generate a suppression instruction based on the wrinkle suppression strategy, and adjust a printing machine unit corresponding to the wrinkle region based on the suppression instruction.

[0007] In a third aspect, the embodiment of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of the above gravure printing wrinkle detection and suppression method when executing the computer program.

[0008] In a fourth aspect, the embodiment of the present application provides a computer readable storage medium, which stores a computer program, wherein the computer program is executed by a processor to implement the steps of the above gravure printing wrinkle detection and suppression method.

[0009] The method and system for detecting and suppressing the wrinkles in the gravure printing provided by the embodiment of the present application have the following advantages: the image acquisition of the full width of the printed paper is realized by the line array camera imaging and image stitching technology, which breaks through the limitations of high missed detection rate and poor timeliness of manual detection. The pixel-level mask output by the pre-trained wrinkle evaluation model solves the industry pain points of fuzzy positioning of the wrinkle area boundary and difficult quantification. Based on the pixel-level analysis of the wrinkle direction and wrinkle density and the division of the wrinkle area into multiple sub-areas, the determination of the wrinkle influence level is realized by calculating the sub-evaluation coefficient through the optical anomaly index and the tension instability index. According to the comprehensive influence level and the multiple sub-areas, a targeted suppression strategy for the wrinkles is formulated and a suppression instruction is generated, which can adjust the operation parameters of the printing machine unit in real time, effectively improve the surface quality and yield of the printed matter, and reduce the frequency of shutdown debugging. The present application improves the automation of the gravure printing production line. BRIEF DESCRIPTION OF DRAWINGS

[0010] Fig. 1 A flowchart of the method for detecting and suppressing the wrinkles in the gravure printing provided by the embodiment of the present application is shown in the figure. Fig. 2 A structure block diagram of the system for detecting and suppressing the wrinkles in the gravure printing provided by the embodiment of the present application is shown in the figure. Fig. 3 A schematic block diagram of the electronic device provided by the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0011] In the following description, specific details are set forth such as particular system configurations, techniques, etc., in order to provide a thorough understanding of the embodiments of the present application. However, it should be apparent to those skilled in the art that the present application can be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.

[0012] In order to make the objects, technical solutions and advantages of the present application clearer, the following will be combined with the accompanying drawings to make a detailed description. Figs. 1-3 The specific embodiments are described.

[0013] Reference should be made to Fig. 1 , Fig. 1 A flowchart of the method for detecting and suppressing the wrinkles in the gravure printing provided by the embodiment of the present application is shown in the figure, which includes: S101: Obtain the line array data of the printed paper by the line array camera and perform stitching to obtain image data; pre-process and extract features from the image data to obtain feature image data.

[0014] In the embodiment, the line array camera has the characteristics of high resolution and high scanning frequency, and can match the transmission speed of the printed paper and collect line array data row by row along the width direction of the paper. Since the single line array data only includes a local narrow area of the paper, the image stitching algorithm is used to align, fuse and stitch the multiple frames of line array data collected continuously according to the spatial correspondence of the pixel coordinates, and finally generate image data including the printed paper surface.

[0015] After obtaining the image data, the embodiment performs preprocessing and feature extraction operations on the image data. For example, the preprocessing of the image data includes image denoising, grayscale, geometric correction and the like. For example, first, the noise points generated by dust and equipment vibration in the printing process are removed by a median filter algorithm, and then the color image is converted into a grayscale image; the image distortion caused by camera installation deviation or printing paper transmission jitter is geometrically corrected to obtain the true form of the paper surface; and the feature extraction is performed on the preprocessed image data. In the embodiment, an edge detection algorithm is used to identify the gray mutation edges of the printed paper surface caused by wrinkles, and a texture feature extraction algorithm is used to analyze the texture difference between the wrinkle area and the flat area, and finally the feature image data highlighting the wrinkle features is obtained.

[0016] S102: input the feature image data into the pre-trained wrinkle evaluation model to obtain the wrinkle area and the pixel-level mask thereof.

[0017] In the embodiment, the feature image data is input into the pre-trained wrinkle evaluation model, and the wrinkle area and the pixel-level mask thereof. The wrinkle evaluation model is constructed based on a deep learning architecture and has the ability of pixel-level feature learning and classification, and can distinguish the wrinkle area and the flat area on the surface of the printed paper.

[0018] When the feature image data is input into the wrinkle evaluation model, the wrinkle evaluation model performs convolution, pooling and up-sampling operations on the image layer by layer, and extracts multi-scale feature information from the shallow layer to the deep layer, wherein the shallow layer feature captures the edge details of the wrinkles, and the deep layer feature represents the overall shape and spatial distribution law of the wrinkles. Secondly, through feature fusion and classifier determination, the wrinkle evaluation model finally outputs two types of results, wherein the first type of result is the overall positioning information of the wrinkle area, such as the specific position and range of the wrinkles on the surface of the paper; and the second type of result is a pixel-level mask, which is presented in the form of a binary image, wherein the value of each pixel point corresponds to whether it belongs to the wrinkle area in the original feature image.

[0019] The wrinkle evaluation model of the embodiment adopts a semantic segmentation deep learning architecture, and the levels include an input layer, a feature extraction layer, a feature fusion layer, and an output layer. The input layer receives feature image data highlighting wrinkle edge and texture difference; the feature extraction layer extracts wrinkle features from shallow edge to deep semantics through convolution layers and pooling layers; the feature fusion layer splices and fuses multi-scale features to strengthen pixel-level detail representation; and the output layer outputs a pixel-level mask consistent with the size of the input image through upsampling and a Sigmoid activation function. The pixel-level mask refers to a pixel image output by the wrinkle evaluation model and consistent with the size of the input feature image, which identifies the wrinkle area and the flat area of the printed paper surface through pixel values 1 and 0.

[0020] Before training the wrinkle evaluation model, a data set is constructed, which is generated by collecting paper images under different printing conditions, pre-processing and feature extraction to generate feature images, and constructing a label with a manually labeled wrinkle area pixel-level mask; in the training process of the wrinkle evaluation model, the feature images and corresponding labels are divided into training data set, validation data set and test data set based on the ratio of 7:2:1, and the cross-entropy loss function is used as the optimization objective, and the small batch gradient descent method is used for iterative training. After each round of training, the segmentation accuracy of the wrinkle evaluation model is evaluated with the validation set, and the training is stopped when the validation set loss value no longer decreases, and finally the generalization ability of the wrinkle evaluation model is verified with the test set, and the wrinkle evaluation model meeting the printing wrinkle detection accuracy requirement is obtained.

[0021] The parameter settings of the wrinkle evaluation model of the embodiment include: the convolution layer uses a 3x3 convolution kernel, the step length is set to 1, and the padding method is SAME padding; the pooling layer uses a 2x2 maximum pooling kernel, and the step length is set to 2; the feature fusion layer uses a transverse splicing method to fuse shallow and deep features; the output layer uses transpose convolution for upsampling, the convolution kernel size is 4x4, and the step length is 2; in terms of training parameters, the batch size is set to 8-16, the initial value of the learning rate is set to 0.001, the Adam optimizer is selected as the optimizer, the weight decay coefficient is set to 0.0001 to prevent overfitting, the number of training rounds is set to 50-100 rounds, and the trigger condition of the early stopping strategy is set to no decrease in the validation set loss for 10 consecutive rounds.

[0022] S103: Analyze the wrinkle direction and wrinkle density of the wrinkle area according to the pixel-level mask, and divide the wrinkle area to obtain a plurality of sub-areas.

[0023] In the embodiment, firstly, it is judged whether the pixel-level mask has a wrinkle direction, and different division methods are selected according to whether the wrinkle direction exists. For example, if the pixel-level mask has a wrinkle direction, the normal direction gradient of each pixel point in the pixel-level mask is calculated, the target wrinkle direction is determined through cluster analysis, then a wrinkle density heat map of the wrinkle region is generated based on the pixel-level mask, and based on the spatial variation characteristics perpendicular to the target wrinkle direction, a plurality of division boundaries are determined, then the wrinkle region is divided based on the plurality of division boundaries to obtain a plurality of strip-shaped sub-regions; if the pixel-level mask does not have a wrinkle direction, a skeletonization algorithm is used to extract the skeleton of the wrinkle in the pixel-level mask, and the branch points and end nodes of the skeleton line are identified, and the branch points and end nodes are taken as vertices, the skeleton path connecting the vertices is taken as a boundary, and the wrinkle region is divided into a plurality of topological polygon sub-regions, and the plurality of topological polygon sub-regions are morphologically optimized, and adjacent sub-regions with an area less than a first preset threshold and / or a shape irregularity greater than a second preset threshold are merged to obtain a plurality of physically merged sub-regions.

[0024] Among them, the plurality of sub-regions include a plurality of physically merged sub-regions and a plurality of strip-shaped sub-regions.

[0025] S104: Calculate the optical property and tension of all sub-regions to obtain an optical property anomaly index and a tension instability index.

[0026] Based on the optical property anomaly index and the tension instability index, a sub-evaluation coefficient of each sub-region is determined.

[0027] According to the distribution characteristics of all sub-evaluation coefficients, a comprehensive influence grade of the wrinkle region is determined.

[0028] In the embodiment, the surface wrinkles of the printed paper will cause uneven physical morphology, under the uniform light source irradiation of the printing detection, the convex parts will form highlight areas due to the change of light reflection angle, and the concave parts will form shadow areas due to light blocking.

[0029] The optical properties and tension of all sub-regions are calculated to obtain the optical anomaly index and tension instability index. Specifically, first, the gray value of the flat paper region is extracted as the reference gray value, and the gray value data of each pixel point in the sub-region is extracted. By comparing the gray value data with the reference gray value of the flat paper region, the pixel points corresponding to the gray values greater than the reference threshold are selected as the highlight pixel group, and the pixel points corresponding to the gray values less than the reference threshold are selected as the shadow pixel group. Then, the area ratio of the highlight region and the shadow region of each sub-region and the gray deviation degree of the highlight and the shadow are calculated and weighted to obtain the optical anomaly index. For example, first, the number of highlight pixel groups in a sub-region, the total number of pixels in the sub-region, the number of shadow pixel groups, and the gray value of each pixel of the highlight pixel group and the shadow pixel group are obtained. The ratio of the number of highlight pixel groups to the total number of pixels in the sub-region is used to obtain the highlight area ratio. The ratio of the number of shadow pixel groups to the total number of pixels in the sub-region is used to obtain the shadow area ratio. Second, the gray deviation degree of the highlight and the shadow is calculated, i.e., the positive deviation of the gray value of each highlight pixel in the highlight pixel group from the reference value is calculated, then the deviation values of all highlight pixels are summed, and finally the total number of highlight pixels is divided to obtain the average gray deviation of the highlight. The negative deviation of the gray value of each shadow pixel in the shadow pixel group from the reference value is calculated, then the deviation values of all shadow pixels are summed, and finally the total number of shadow pixels is divided to obtain the average gray deviation of the shadow.

[0030] The four parameters are combined based on a preset weight to obtain the optical anomaly index of the region, wherein the higher the value of the optical anomaly index, the more serious the optical distortion caused by wrinkling. The weight combination is set according to the difference in tolerance of the printed matter to highlight defects and shadow defects, and according to the collection of normal samples and typical wrinkling samples in the history production, the highest correlation between the optical anomaly index of the sub-region and the artificial labeled wrinkling severity score is used as the optimization target, and the weight combination is adjusted by using multivariate linear regression or grid search method, so as to obtain a set of weight values suitable for the current production line materials and process characteristics.

[0031] For each sub-region, the edge profile data of the printed pattern is extracted, the actual edge coordinates of the pattern are obtained through the edge detection algorithm, and the standard edge coordinates in the wrinkle-free state are compared to calculate the edge offset and the distortion degree; at the same time, the edge sharpness evaluation function is used to quantify the blur degree of the pattern edge. After normalization processing of the edge offset, distortion degree and blur degree, the tension instability index of the sub-region is obtained by weighted fusion, wherein the higher the tension instability index value, the more serious the tension imbalance degree of the region. The fusion weights of the edge offset, distortion degree and blur degree are determined based on the contribution analysis of the three indexes to the tension imbalance, the quality priority of the printing process and the quantitative verification of the historical data.

[0032] After the optical anomaly index and the tension instability index of all sub-regions, a weighted evaluation model is constructed, and the weight coefficients of the two indexes are set based on the printing process requirements. For example, in the printing scene with color restoration requirement, the weight of the optical anomaly index is adjusted; in the scene with registration accuracy requirement, the weight of the tension instability index is adjusted. After multiplying the two indexes by the corresponding weights and summing, the sub-evaluation coefficient of each sub-region is obtained. The construction of the weighted evaluation model aims to quantify the wrinkle influence degree of the sub-region, dynamically allocates the weight based on the process requirement difference, and the input of the weighted evaluation model is two core quantitative indexes: the first index is the optical anomaly index representing the wrinkle optical distortion degree, and the second index is the tension instability index representing the pattern deformation degree caused by tension imbalance; the output of the weighted evaluation model is the sub-evaluation coefficient of a single sub-region, which is used to represent the influence weight of the wrinkle of the sub-region on the printing quality. The weighted evaluation model adopts a linear weighted summation model as the basic framework, and the model formula is: y = W1 x I1 + W2 x I2, wherein y is the sub-evaluation coefficient; W1 is the weight coefficient of the optical anomaly index; W2 is the weight coefficient of the tension instability index, and satisfies W1 + W2 = 1; I1 is the optical anomaly index; I2 is the tension instability index.

[0033] The specific value of the weight coefficient is optimized according to the correlation between the indexes and the quality defects of the printed matter in the historical production data, and through multivariate linear regression analysis. Among them, the basic value of W1 and W2 is set based on the general quality priority of the printing process, the statistical data of historical defects and the influence analysis of the two indexes on the wrinkle.

[0034] After obtaining the sub-evaluation coefficients of all sub-regions, the comprehensive influence level of the wrinkled region is determined based on the numerical distribution characteristics and spatial distribution characteristics of all sub-evaluation coefficients. Specifically, the coefficient mean, coefficient variance and coefficient peak proportion of the sub-evaluation coefficient are calculated, and then the comprehensive influence level is matched based on the gravure printing wrinkling comprehensive influence level division rule table in Table 1. Among them, when the coefficient mean is in the moderate interval, but the coefficient peak proportion is in the severe interval, the wrinkled region is determined to be a severe influence level according to the principle of not being low; when the coefficient mean is in the mild interval, but the coefficient variance reaches the moderate interval, the wrinkled region is determined to be a moderate influence level.

[0035] Table 1 Gravure printing wrinkling comprehensive influence level division rule table Overall impact rating Coefficient mean threshold interval Coefficient variance threshold interval Coefficient peak proportion threshold interval Mild <0.3 <0.05 =0 Moderate 0.3 < mean < 0.7 0.05 < variance < 0.2 > 0 and < 20% Severe ≥0.7 ≥0.2 ≥20% S105: Determine the wrinkling suppression strategy based on the comprehensive influence level and the plurality of sub-regions.

[0036] In this embodiment, the type of the wrinkling suppression strategy is determined based on the plurality of sub-regions, and a preliminary suppression strategy is obtained. For example, if the plurality of sub-regions are a plurality of belt-shaped sub-regions, the preliminary suppression strategy is determined to be a tension gradient adjustment strategy; if the plurality of sub-regions are a plurality of physically merged sub-regions, the preliminary suppression strategy is determined to be a unit cooperative adjustment strategy; if the wrinkled region has both belt-shaped sub-regions and physically merged sub-regions, the wrinkled region is determined to be a mixed type of wrinkling, and the preliminary suppression strategy is determined according to the projection distribution of all sub-regions in the paper running direction. Specifically, if the belt-shaped sub-regions dominate in the projection distribution, the preliminary suppression strategy is dominated by the tension gradient adjustment strategy, and the unit cooperative adjustment strategy is superimposed; if the physically merged sub-regions dominate in the projection distribution, the preliminary suppression strategy is dominated by the unit cooperative adjustment strategy, and the tension gradient adjustment strategy is superimposed.

[0037] After obtaining the preliminary suppression strategy, the influence level-basic adjustment amount mapping table is queried based on the comprehensive influence level to obtain the corresponding basic adjustment amount of the printing unit. Based on the basic adjustment amount, the preliminary suppression strategy, the distribution position of the plurality of sub-regions and the sub-evaluation coefficient, the spatial distribution weighting and priority allocation are performed on the basic adjustment amount to obtain the wrinkling suppression strategy.

[0038] S106: Generate a suppression instruction based on the wrinkling suppression strategy, and adjust the printing unit corresponding to the wrinkled region based on the suppression instruction.

[0039] In the embodiment, the parameter adjustment amount in the wrinkle suppression strategy is analyzed and verified with the operating state data, and the wrinkle suppression strategy that passes the verification is resolved and quantized into a series of atomized low-level driving commands with time sequence logic to form standardized control instructions according to the priority of the defined parameters, the basic adjustment amount and the corresponding printing unit, and based on the state parameters of each adjustment unit obtained from the printing machine control synchronization.

[0040] The suppression instruction is a driving command that can be recognized and executed by the distributed control system of the printing machine, and the printing unit corresponding to the wrinkled region is one or more specific printing execution mechanisms that can effectively suppress wrinkles.

[0041] From the above, the present application realizes image acquisition of the full width of the printed paper through line array camera imaging and image stitching technology, breaking through the limitations of high missed detection rate and poor timeliness of manual detection. The pre-trained wrinkle evaluation model outputs a pixel-level mask, solving the industry pain points of fuzzy positioning of the wrinkled region boundary and difficulty in quantification. Based on pixel-level analysis of the wrinkled direction and wrinkle density, the wrinkled region is divided into multiple sub-regions, and then the sub-evaluation coefficients are calculated based on the optical anomaly index and the tension instability index, realizing the determination of the wrinkle influence grade. According to the comprehensive influence grade and multiple sub-regions, a specific wrinkle suppression strategy is developed and a suppression instruction is generated, which can adjust the operating parameters of the printing unit in real time, effectively improve the surface quality of the printed product and the yield, and reduce the frequency of shutdown debugging. The present application improves the automation and intelligence level of the gravure printing production line.

[0042] In an embodiment of the present application, the wrinkled direction and wrinkle density of the wrinkled region are analyzed according to the pixel-level mask, and the wrinkled region is divided to obtain multiple sub-regions, including: determining whether the pixel-level mask has a wrinkled direction; if there is a wrinkled direction, calculating the normal direction gradient of each pixel point in the pixel-level mask, and determining the target wrinkled direction through clustering analysis; generating a wrinkle density heat map of the wrinkled region based on the pixel-level mask; determining multiple segmentation boundaries based on the spatial variation characteristics perpendicular to the target wrinkled direction of the heat map; segmenting the wrinkled region based on the multiple segmentation boundaries to obtain multiple strip-shaped sub-regions; if there is no wrinkled direction, executing a topological structure division sub-mechanism to obtain multiple physically merged sub-regions; The plurality of sub-regions include a plurality of physically merged sub-regions and a plurality of strip-shaped sub-regions.

[0043] In the embodiment, the pixel-level mask is subjected to direction feature recognition, and it is judged whether the pixel-level mask has a wrinkle direction based on the recognized direction feature. If the wrinkle direction exists, the strip-shaped sub-region division process is started; if the wrinkle direction does not exist, that is, the wrinkle morphology presents a random and diffuse distribution, the topological structure division sub-mechanism is switched to.

[0044] For example, if it is determined that the pixel-level mask has a wrinkle direction, the target extension direction is further obtained through normal direction gradient calculation and clustering analysis. Specifically, first, the wrinkle region marked in the pixel-level mask is taken as an analysis object, and the normal direction gradient of each wrinkle pixel is calculated point by point. The normal direction gradient is a vector parameter, which has two core attributes of direction and amplitude: the direction is perpendicular to the tangent direction of the wrinkle edge where the pixel point is located, and points to the extension direction of the wrinkle; the amplitude, that is, the gradient value, is positively correlated with the gray difference of the pixel point and the neighborhood pixels, wherein the higher the amplitude, the more obvious the wrinkle protrusion or depression at the position, and the clearer the edge contour. By traversing all the pixels in the wrinkle region point by point, a complete normal direction gradient dataset including direction and amplitude information is constructed. Subsequently, the K-means clustering algorithm is used to group process the above gradient dataset, and the preset number of clustering clusters is 3-5 to adapt to wrinkle morphologies of different complexity. The K-means clustering algorithm takes the similarity of gradient direction as the core clustering basis, and takes the amplitude as the auxiliary screening condition to remove low-confidence gradient data with an amplitude less than a preset low-confidence threshold, thereby reducing the interference of image noise and edge burrs on the clustering result. At the same time, the center positions of each clustering cluster are iteratively optimized to minimize the variance of the gradient direction within the cluster, and finally the pixel points with similar gradient directions are divided into the same clustering cluster to obtain the clustering result. The cluster center corresponding to the gradient direction of the cluster with the highest pixel ratio in the clustering result is determined as the target wrinkle direction. The preset low-confidence threshold is set based on the image noise level of the pixel-level mask, the statistical distribution of the gradient amplitude of the wrinkle edge, and the target accuracy of the clustering analysis. Secondly, a wrinkle density heat map of the wrinkle region is generated based on the pixel-level mask. Specifically, by using a preset sliding window, the entire wrinkle region is traversed, and the proportion of wrinkle pixels in each sliding window is calculated as the wrinkle density, and the wrinkle density is visualized and mapped by color depth, wherein the sliding window is determined based on the typical dimensional features of the printed wrinkle region, the spatial resolution requirement of the wrinkle density, and the calculation efficiency.

[0045] In this embodiment, the segmentation boundary is determined based on the spatial variation characteristics of the heat map in the direction perpendicular to the wrinkle direction of the target. Since the extension direction of the strip-shaped sub-region is consistent with the wrinkle direction of the target, the direction perpendicular to the wrinkle direction of the target is the sensitive direction of the density variation. The mutation points of the density in the sensitive direction in the heat map are extracted as the segmentation boundary points, and the continuous segmentation boundary points are fitted as a plurality of parallel segmentation boundaries. The wrinkle region is longitudinally cut to obtain a plurality of strip-shaped sub-regions.

[0046] If it is determined that the pixel-level mask does not have a wrinkle direction, a topological structure division sub-mechanism is performed. The topological structure division sub-mechanism takes the topological connectivity of the wrinkle region as the division basis, extracts independent connected domains in the pixel-level mask, and regards each connected domain as an independent physical merging sub-region to obtain a plurality of physical merging sub-regions with independent boundaries and complete topological structures.

[0047] From the above, it can be seen that the embodiment first determines whether the pixel-level mask has a wrinkle direction, and then uses a differentiated region division strategy accordingly, thereby improving the adaptability to different forms of wrinkle characteristics. The targeted division strategy ensures the internal consistency of the sub-regions after division and realizes the full coverage of complex wrinkle scenes, thereby improving the targeting and efficiency of wrinkle suppression.

[0048] In an embodiment of the present application, if there is no wrinkle direction, a topological structure division sub-mechanism is performed to obtain a plurality of physical merging sub-regions, including: The skeletonization algorithm is used to extract the skeleton of the wrinkle in the pixel-level mask, and the branch points and end nodes of the skeleton line are identified. The branch points and end nodes are taken as vertices, the skeleton path connecting the vertices is taken as a boundary, and the wrinkle region is divided into a plurality of topological polygon sub-regions.

[0049] The morphological optimization is performed on the plurality of topological polygon sub-regions, the adjacent sub-regions with an area less than a preset first threshold value and / or a shape irregularity greater than a preset second threshold value are merged, and a plurality of physical merging sub-regions are obtained.

[0050] In the embodiment, firstly, the wrinkled region in the pixel-level mask is processed by using a skeletonization algorithm. The skeletonization algorithm can gradually erode the edge pixels of the wrinkled region while preserving the morphological characteristics of the wrinkles, and finally extract a single-pixel-width skeleton of the wrinkles to represent the overall topological connectivity structure of the wrinkles. Then, the skeleton line is extracted and feature point identification is performed to screen out branch points where the skeleton lines intersect and end nodes at the ends of the skeleton lines. Secondly, the branch points and end nodes are taken as the vertices of the topological polygon, and the skeleton path between adjacent vertices is taken as the boundary of the polygon. By polygon fitting, the originally connected and diffuse wrinkled region is divided into multiple topological polygon sub-regions. Finally, morphological optimization is performed on the multiple topological polygon sub-regions. Specifically, two quantitative screening thresholds are preset: the first threshold for representing the size of the sub-region, and the second threshold for evaluating the morphological regularity of the sub-region. The second threshold is the shape irregularity threshold, which is determined by calculating parameters such as the ratio of the perimeter to the area of the sub-region. The first threshold is determined by first collecting area samples of the topological polygon sub-regions of the diffuse wrinkled cases, removing outliers, and then statistically analyzing the minimum effective area interval. Based on the adaptability experiment of the wrinkled suppression strategy after sub-region merging, an area threshold is selected that can avoid interference with the analysis accuracy of fine sub-regions while preserving the key morphological characteristics of the wrinkled region. This threshold is the first threshold. For small sub-regions with an area less than or equal to the first threshold and / or deformed sub-regions with a shape irregularity greater than or equal to the second threshold, they are merged with adjacent sub-regions with similar features, and finally multiple physically merged sub-regions are obtained.

[0051] The input of the skeletonization algorithm is the wrinkled region in the pixel-level mask, and the output is a single-pixel-width skeleton line that preserves the morphological characteristics of the wrinkles, as well as branch points and end nodes identified therefrom. The single-pixel-width skeleton line simplifies the two-dimensional diffuse wrinkled region to a one-dimensional connected line, preserving the topology of the wrinkles such as direction and connectivity. The branch points correspond to the branching positions of the wrinkles, and the end nodes correspond to the starting or ending positions of the wrinkles. These feature points are the vertices for dividing the topological polygon sub-regions.

[0052] The parameters of the skeletonization algorithm are set based on the accuracy and practicality requirements of industrial wrinkle analysis. The parameters include the erosion termination parameter, the feature point identification parameter, and the noise filtering parameter. The erosion termination parameter is used to control the number of iterations. The feature point identification parameter is used to distinguish branch points from end nodes by using a neighborhood pixel threshold. The noise filtering parameter is used to remove isolated skeleton segments that are too short and reduce the interference of fine features on subsequent sub-region division.

[0053] From the above, the embodiment extracts the wrinkle skeleton by the skeleton algorithm, identifies the branch points and end nodes, and divides the topological polygon sub-regions in combination with the skeleton path. Then, the small or deformed sub-regions are merged through morphological optimization, which not only preserves the topological structure characteristics of the irregular wrinkle region, but also ensures the morphological regularity and scale rationality of the physical merged sub-region. The scientific and reliable partition basis is provided for the diffuse wrinkle scene, and the adaptability and execution efficiency of the wrinkle suppression strategy are improved.

[0054] In an embodiment of the present application, a gravure printing wrinkle detection and suppression method further comprises: calculating the sub-region number, average area and average shape irregularity of the plurality of physical merged sub-regions; comparing the sub-region number and average shape irregularity with the preset target number range and target average irregularity; if the sub-region number is greater than the upper limit of the target number range, using a preset first step to increase the first threshold value; if the sub-region number is less than the lower limit of the target number range, using the first step to decrease the first threshold value; if the average shape irregularity is greater than the target average irregularity, using a preset second step to increase the second threshold value; calculating the absolute value of the difference between the average area and the preset target average area to obtain the average area deviation; if the average area deviation is less than or equal to a preset third threshold value, continue to adjust the first threshold value and the second threshold value until the sub-region number and the average shape irregularity simultaneously satisfy the target number range and the target average irregularity; if the average area deviation is greater than the third threshold value, trigger an adjustment process abnormal warning or reset the first threshold value.

[0055] In the embodiment, first, the quantitative feature statistics of all physical merged sub-regions are performed to obtain the number of physical merged sub-regions, calculate the average area and average shape irregularity of the number of physical merged sub-regions. Then, the number of physical merged sub-regions and the average shape irregularity are compared with the preset target number range and target average irregularity, respectively, to obtain the comparison result.

[0056] Subsequently, a differential threshold adjustment operation is performed based on the comparison result. Specifically, if the number of physical merged sub-regions is greater than the upper limit of the target number range, it indicates that there are too many fine sub-regions, in which case the first threshold is increased by a preset first step to reduce the number of physical merged sub-regions; if the number of physical merged sub-regions is less than the lower limit of the target number range, it indicates that the physical merged sub-regions are excessively merged, in which case the first threshold is decreased by the first step to split more physical merged sub-regions; if the average shape irregularity is greater than the target average irregularity, it indicates that the physical merged sub-regions are not regular enough in shape, in which case the second threshold is increased by a preset second step, wherein the first step and the second step are dynamically optimized based on the printing process requirements, equipment adjustment precision, historical iteration optimization data, and sub-region division target using an adaptive gradient descent algorithm based on momentum acceleration, and numerical boundary constraints are applied to keep them within a preset effective value range. The target number range is a reasonable interval for the number of physical merged sub-regions, and the determination of the upper and lower limits anchors two core dimensions: one is the implementation accuracy of the wrinkle suppression strategy, and the other is the efficiency requirement of industrial production. The target average irregularity is a critical indicator of the shape regularity of the physical merged sub-regions, and its value is determined based on the subsequent analysis of the matching sub-regions and the application scenario requirements.

[0057] Further, in the adjustment process of the first threshold and the second threshold, the absolute value of the difference between the average area of the physical merged sub-regions and the preset target average area is calculated to obtain the average area deviation. For example, if the average area deviation is less than or equal to a preset third threshold, it indicates that the size of the physical merged sub-regions is within an acceptable range, and the first threshold and the second threshold continue to be adjusted until the number of physical merged sub-regions and the average shape irregularity meet the preset target requirements; if the average area deviation is greater than the third threshold, it indicates that the size of the physical merged sub-regions is in an unacceptable range, and an abnormal warning is triggered to remind the operator to intervene in checking the setting logic of the first threshold and the second threshold, the execution state of the sub-region division algorithm, etc., or to perform a first threshold reset operation to restart the whole-process sub-region division and threshold iteration work. The third threshold is calculated based on the process adaptation area of the physical merged sub-regions, the statistical deviation range of the historical division data, and the target precision requirements of wrinkle suppression.

[0058] In this embodiment, the adaptive gradient descent algorithm based on momentum acceleration is used to adjust the first step size and the second step size. Specifically, the momentum cache of the first step size and the second step size is initialized to zero; at each adjustment of the first threshold and the second threshold, the prediction error change rate of the number of sub-regions and the prediction error change rate of the average shape irregularity in the current iteration round are calculated; the first step size is updated according to the prediction error change rate of the number of sub-regions and the momentum cache of the previous iteration, and the momentum cache of the first step size is updated synchronously; the update direction of the first step size is the direction that makes the error of the number of sub-regions decrease fastest; the second step size is updated according to the prediction error change rate of the average shape irregularity and the momentum cache of the previous iteration, and the momentum cache of the second step size is updated synchronously; the update direction of the second step size is the direction that makes the error of the average shape irregularity decrease fastest; numerical boundary constraints are applied to the updated first step size and the second step size to keep them within the preset effective range. The preset effective value range is the upper and lower limit interval of the step size determined based on the printing process parameters, the equipment adjustment limit, the historical iteration optimization data and the sub-region division target accuracy.

[0059] The adaptive gradient descent algorithm of this embodiment is an optimization algorithm based on momentum acceleration mechanism. Its core is to incorporate the momentum information of historical iterations into the gradient descent (iterative optimization along the direction of fastest error reduction) to make the parameter update process have inertia, which can not only speed up the convergence, but also reduce the oscillation in the iteration process, and is often used for dynamic adjustment of parameters. The momentum cache is used to store the momentum information of the step size update in the previous iteration process, and its initial value is set to 0. Its role is to make the update of the current step size not only refer to the error change rate of the current round, but also integrate the momentum trend of historical iterations; the prediction error change rate of the number of sub-regions refers to the deviation of the actual number of sub-regions from the target number range in the current iteration round, and the change amplitude and trend of the deviation relative to the last iteration, which represents the change speed and direction of the number of sub-regions error, and is the basis for updating the first step size; the prediction error change rate of the average shape irregularity refers to the deviation of the actual average shape irregularity from the target value in the current iteration round, and the change amplitude and trend of the deviation relative to the last iteration, which represents the change speed and direction of the shape regularity error, and is the basis for updating the second step size; the numerical boundary constraint is the effective value range set for the updated first step size and the second step size, and its purpose is to avoid the step size being too large to cause the threshold adjustment amplitude to be too large; at the same time, it avoids the step size being too small to cause the iteration to converge too slowly.

[0060] From the above, it can be seen that this embodiment statistically counts the number, average area and average shape irregularity of the physical merged sub-regions, and compares them with the preset target values, and increases or decreases the first and second thresholds accordingly, and at the same time, uses the average area deviation as the basis for abnormality judgment, which not only improves the control accuracy of the number and shape of the physical merged sub-regions, but also ensures the timely identification of abnormal situations.

[0061] In an embodiment of the present application, the suppression creping strategy is determined based on the comprehensive influence level and the plurality of sub-regions, comprising: determining the type of the suppression creping strategy based on the plurality of sub-regions to obtain a preliminary suppression strategy; querying a preset influence level-basic adjustment amount mapping table based on the comprehensive influence level to obtain the basic adjustment amount of the printing unit corresponding to the creping region; based on the basic adjustment amount, the preliminary suppression strategy, the distribution position of the plurality of sub-regions and the sub-evaluation coefficient, performing spatial distribution weighting and priority allocation on the basic adjustment amount to obtain the suppression creping strategy.

[0062] In the embodiment, first, the core type of the suppression strategy is determined based on the plurality of sub-regions obtained by division to obtain a preliminary suppression strategy. Specifically, for the strip-shaped sub-region extending along the target creping direction, the preliminary strategy is set to be the gradient tension control along the vertical creping direction; for the irregularly shaped physically merged sub-region, the preliminary strategy is set to be the core direction of local region pressure and tension coordinated adjustment.

[0063] Subsequently, according to the comprehensive influence level of the creping region, a preset influence level-basic adjustment amount mapping table is queried to obtain the printing unit basic adjustment amount matched with the comprehensive influence level. The mapping table is constructed based on a large amount of process data and experimental results.

[0064] Finally, taking the queried printing unit basic adjustment amount as the benchmark, and based on the direction of the preliminary suppression strategy control, while taking into account the distribution position difference of the plurality of sub-regions on the printing web and the size of the sub-evaluation coefficient of each sub-region, the spatial distribution weighting and priority allocation operation is performed on the basic adjustment amount, and finally the suppression creping strategy is formed which takes into account the global control demand and the local precise adaptation. Among them, the basic adjustment amount includes the basic adjustment amount of tension, the basic adjustment amount of pressure roller gap parameter, the basic adjustment amount of paper conveying speed, the basic adjustment amount of motor load rate, etc.

[0065] For example, a certain web press has a diffuse wrinkle on the printing substrate web, which is divided into 5 physically merged sub-regions. After querying the mapping table, it is determined that the basic tension adjustment amount is increased by 5N, and the preliminary suppression strategy is local pressure and tension coordinated adjustment. Sub-regions A and B are located in the printing core section, and the sub-evaluation coefficients are 0.9 and 0.85 (greater than the threshold value 0.7); sub-regions C, D, and E are located in the non-core section of the edge section, and the sub-evaluation coefficients are 0.5, 0.4, and 0.35 (less than the threshold value 0.7). After weighted distribution is performed, the tension of A and B is increased to 8N and is matched with a small pressure reduction, and the tension of C, D, and E is only increased by 3N, and finally a suppression strategy is formed which takes into account the key adjustment and control of the core region and the moderate adjustment of the edge region. The threshold value 0.7 is a judgment threshold value determined by combining data analysis and experimental verification based on historical process data of printing production, wrinkle suppression effect statistics, and target control accuracy.

[0066] From the above, it can be concluded that the embodiment determines the preliminary suppression direction through sub-region orientation, and matches the basic adjustment amount based on the comprehensive influence level, and then performs weighted distribution based on the sub-region distribution position and the sub-evaluation coefficient, which not only ensures the control strength of the core region, but also avoids excessive adjustment of the non-critical region, effectively improving the pertinence and overall efficiency of wrinkle suppression.

[0067] In an embodiment of the present application, the type of the suppression wrinkle strategy is determined based on a plurality of sub-regions to obtain a preliminary suppression strategy, including: If the plurality of sub-regions are a plurality of strip-shaped sub-regions, the preliminary suppression strategy is determined to be a tension gradient adjustment strategy; If the plurality of sub-regions are a plurality of physically merged sub-regions, the preliminary suppression strategy is determined to be a unit coordinated adjustment strategy; If the wrinkle region has both strip-shaped sub-regions and physically merged sub-regions, the wrinkle region is determined to be a mixed type of wrinkle, and the preliminary suppression strategy is determined according to the projection distribution of all sub-regions in the paper running direction; If the strip-shaped sub-regions dominate in the projection distribution, the preliminary suppression strategy is dominated by the tension gradient adjustment strategy, with the unit coordinated adjustment strategy superimposed; If the physically merged sub-regions dominate in the projection distribution, the preliminary suppression strategy is dominated by the unit coordinated adjustment strategy, with the tension gradient adjustment strategy superimposed.

[0068] In the embodiment, if the wrinkle region is divided into only a plurality of strip-shaped sub-regions, the preliminary suppression strategy is determined to be a tension gradient adjustment strategy. The logic of this tension gradient adjustment strategy is to set a gradientized tension adjustment parameter along the dimension perpendicular to the target wrinkle direction, and to eliminate the extension trend of the strip-shaped wrinkle through a differentiated tension distribution, which adapts to the directional distribution characteristics of the strip-shaped sub-region.

[0069] If only multiple physically merged sub-regions are obtained after the wrinkling region is divided, the preliminary suppression strategy is determined as a unit coordination adjustment strategy. The unit coordination adjustment strategy focuses on the coordination of each execution unit of the printing machine, solves the irregular wrinkling problem of the diffused physically merged sub-region by synchronously adjusting the tension and pressure parameters of the local region, and realizes accurate control of local wrinkling.

[0070] If the wrinkling region includes both the strip-shaped sub-region and the physically merged sub-region, it is determined that the region is mixed wrinkling, and the dominant direction of the preliminary suppression strategy is further determined according to the projection distribution proportion of all sub-regions in the paper running direction. If the projection area proportion of the strip-shaped sub-region is greater than the projection area proportion of the physically merged sub-region, it indicates that the projection area proportion of the strip-shaped sub-region is higher and occupies a dominant position in the distribution, so the preliminary suppression strategy is dominated by the tension gradient adjustment strategy, and the unit coordination adjustment strategy is superimposed to consider the local regulation of the physically merged sub-region. If the projection area proportion of the physically merged sub-region is greater than the projection area proportion of the strip-shaped sub-region, it indicates that the projection area proportion of the physically merged sub-region is higher and occupies a dominant position in the distribution, so the preliminary suppression strategy is dominated by the unit coordination adjustment strategy, and the tension gradient adjustment strategy is superimposed to ensure the directional regulation demand of the strip-shaped sub-region.

[0071] From the above, it can be concluded that the embodiment matches the wrinkling suppression strategy according to the type difference of the sub-region, uses the tension gradient adjustment or unit coordination adjustment strategy for single type sub-region, and determines the main and auxiliary strategies based on the projection distribution proportion of the sub-region for mixed wrinkling and superimposes and implements them, which improves the adaptability to different wrinkling scenarios, and also effectively improves the scientificity and pertinence of the wrinkling suppression strategy.

[0072] In an embodiment of the present application, before generating the suppression instruction based on the wrinkling suppression strategy and adjusting the printing machine unit corresponding to the wrinkling region based on the suppression instruction, it further includes: Obtain the running state data of the printing machine unit corresponding to the wrinkling region; Co-analyze and safety boundary check the parameter adjustment amount in the wrinkling suppression strategy and the running state data to obtain an analysis and check result; Optimize and conflict resolution the parameter adjustment amount based on the analysis and check result to obtain the wrinkling suppression strategy for generating the suppression instruction.

[0073] In the embodiment, first, the printing machine unit running state data directly associated with the wrinkling region is obtained, and the running state data includes the tension real-time value, the pressure roller gap parameter, the paper conveying speed, the motor load rate, etc.

[0074] Secondly, the basic adjustment amount in the wrinkle suppression strategy is analyzed and verified in coordination with the collected real-time running state data. The coordination analysis mainly verifies the matching degree of the parameter adjustment amount and the current running condition, and judges whether the adjustment action conforms to the logical correlation of the printing process. The safety boundary verification part verifies whether the parameter adjustment is out of the safe running interval of the equipment according to the pre-set equipment safety running threshold, including the upper limit of tension adjustment, the adjustment range of pressure roller gap, etc. Finally, the analysis and verification results including matching degree evaluation and safety risk judgment are formed. The equipment safety running threshold is the critical interval of the equipment running parameter determined according to the hardware performance parameters of the gravure printing machine, the standard requirements of the printing process, the historical fault and process optimization data, and the verification experiment of multiple working conditions.

[0075] Finally, based on the above analysis and verification results, the basic adjustment amount in the original wrinkle suppression strategy is optimized and conflict is resolved. For the parameters with insufficient matching degree, fine adjustment is carried out based on the real-time working condition of the unit; for the adjustment amount exceeding the safety boundary, it is corrected to the safety interval; for the logical conflict between different parameters, the core wrinkle suppression target is prioritized and the conflict is resolved, and finally a set of wrinkle suppression strategy with safety, rationality and executability is obtained, which is used to generate and issue the formal suppression instructions to the printing machine unit.

[0076] After obtaining the wrinkle suppression strategy, the embodiment can also determine whether to trigger the multi-unit linkage suppression mode according to the starting position, range and comprehensive influence level of the wrinkle area. For example, if the multi-unit linkage suppression mode is triggered, based on the wrinkle suppression strategy, a first adjustment instruction set for the current unit corresponding to the wrinkle area and a second adjustment instruction set for at least one associated unit located upstream of the current unit are generated; the current unit and the associated unit are controlled to adjust in coordination based on their respective adjustment instruction sets.

[0077] The multi-unit linkage suppression mode of the embodiment refers to the control mode in which the unit currently having wrinkle problem and at least one associated unit upstream thereof cooperate and synchronously execute adjustment actions in the gravure printing production line; the current unit refers to the printing machine unit located upstream of the current unit and directly related to the paper conveying path.

[0078] The first adjustment instruction set is a set of adjustment instructions generated for the current unit, which is used to execute direct suppression actions on the wrinkle area. The instruction content includes parameters related to the current unit, such as adjusting the coating pressure, local tension compensation, adjusting the doctor blade angle, etc., which directly act on the work station where the wrinkle occurs to eliminate the wrinkle.

[0079] The second adjustment instruction set refers to a set of adjustment instructions specially generated for the upstream associated group, which is used to perform pre-adjustment actions before the paper raw material in the creping area enters the current group. The instruction content includes eliminating or weakening the transitive process state leading to creping, such as adjusting the unwinding speed of the upstream unwinding group, optimizing the preset value of the tension adjusting group, calibrating the parallelism of the paper guide roller, etc.

[0080] From the above, it can be concluded that the embodiment avoids the problems of adjustment failure caused by the mismatch between the basic adjustment amount and the current working condition of the group, and eliminates the risk of equipment failure or printing defect caused by the basic adjustment amount exceeding the safe range of the equipment, further improves the overall effect of creping suppression and the stability of the production process.

[0081] Corresponding to the gravure printing creping detection and suppression method of the above embodiment, Fig. 2 The structure block diagram of the gravure printing creping detection and suppression system provided by an embodiment of the present application is shown. For the sake of convenience, only the parts related to the embodiments of the present application are shown. For reference Fig. 2 The gravure printing creping detection and suppression system 20 includes an image acquisition and processing module 21, a creping area determination module 22, a creping area division module 23, an influence level determination module 24, a suppression strategy determination module 25, and an instruction generation and execution module 26.

[0082] The image acquisition and processing module 21 is used to acquire line array data of printed paper by a line array camera and splice them to obtain image data; and pre-process and feature extract the image data to obtain feature image data; The creping area determination module 22 is used to input the feature image data into a pre-trained creping evaluation model to obtain the creping area and its pixel-level mask; The creping area division module 23 is used to analyze the creping direction and wrinkle density of the creping area according to the pixel-level mask, and divide the creping area to obtain a plurality of sub-areas; The influence level determination module 24 is used to calculate the optical property abnormality index and the tension instability index of all sub-areas; determine the sub-evaluation coefficient of each sub-area based on the optical property abnormality index and the tension instability index; and determine the comprehensive influence level of the creping area according to the distribution characteristics of all sub-evaluation coefficients; The suppression strategy determination module 25 is used to determine the creping suppression strategy based on the comprehensive influence level and the plurality of sub-areas; The instruction generation and execution module 26 is used to generate suppression instructions based on the creping suppression strategy, and adjust the printing machine unit corresponding to the creping area based on the suppression instructions.

[0083] See Fig. 3 , Fig. 3 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Fig. 3 The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of the modules in the aforementioned device embodiments, for example... Fig. 2 The functions of the image acquisition and processing module 21, the wrinkled area determination module 22, the wrinkled area division module 23, the influence level determination module 24, the suppression strategy determination module 25, and the instruction generation and execution module 26 are shown.

[0084] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0085] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0086] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory. For example, the memory 304 may also store device type information.

[0087] In a particular implementation, the processor 301, the input device 302, and the output device 303 described in the embodiments of the present application can perform the implementation described in any of the embodiments of the gravure printing wrinkle detection and suppression method provided by the embodiments of the present application, and can also perform the implementation of the electronic device described in the embodiments of the present application, which will not be described here.

[0088] In another embodiment of the present application, a computer readable storage medium is provided, which stores a computer program. The computer program includes program instructions, which are executed by a processor to implement all or part of the processes of the above-mentioned embodiments. The computer program can also be used to instruct related hardware to complete the implementation. The computer program can be stored in a computer readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned method embodiments can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0089] The computer readable storage medium can be an internal storage unit of the electronic device of any of the above-mentioned embodiments, such as a hard disk or a memory of the electronic device. The computer readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the computer readable storage medium can include both the internal storage unit and the external storage device of the electronic device. The computer readable storage medium is used to store the computer program and other programs and data required by the electronic device. The computer readable storage medium can also be used to temporarily store data that has been output or will be output.

[0090] Those skilled in the art can clearly understand that, for the convenience and conciseness of description, the specific working process of the electronic device and the units described above can refer to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0091] Those skilled in the art can clearly understand that, for the convenience and conciseness of description, the specific working process of the electronic device and the units described above can refer to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0092] In several embodiments provided in the present application, it should be understood that the disclosed electronic device and method can be implemented in other ways. For example, the above-described device embodiments are only schematic, for example, the division of units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface or unit, and can also be electrical, mechanical or other forms of connection.

[0093] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiments of the present application.

[0094] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0095] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method of detecting and suppressing scudding in gravure printing, characterized by, The method comprises the following steps: acquiring line array data of printed paper by a line array camera and stitching to obtain image data; preprocessing and feature extraction are performed on the image data to obtain feature image data; the feature image data is input into a pre-trained creping evaluation model to obtain a creping area and a pixel-level mask; the creping direction and the wrinkle density of the creping area are analyzed according to the pixel-level mask, and the creping area is divided to obtain a plurality of sub-areas; the optical property and the tension of all sub-areas are calculated to obtain the optical property abnormality index and the tension instability index corresponding to each sub-area; based on the optical property abnormality index and the tension instability index corresponding to each sub-area, a sub-evaluation coefficient of each sub-area is determined; based on the distribution characteristics of all sub-evaluation coefficients, a comprehensive influence level of the creping area is determined; based on the comprehensive influence level and the plurality of sub-areas, a creping inhibition strategy is determined; based on the creping inhibition strategy, a suppression instruction is generated, and the corresponding printing machine unit of the creping area is adjusted based on the suppression instruction.

2. A method of detecting and suppressing scudding in gravure printing according to claim 1, wherein The analysis of the creping direction and the wrinkle density of the creping area according to the pixel-level mask and the division of the creping area to obtain a plurality of sub-areas comprise: determining whether the pixel-level mask has a creping direction; if the creping direction exists, calculating the normal direction gradient of each pixel point in the pixel-level mask, and determining the target creping direction through cluster analysis; generating a wrinkle density heat map of the creping area based on the pixel-level mask; based on the spatial variation characteristics perpendicular to the target creping direction of the heat map, a plurality of segmentation boundaries are determined; based on the plurality of segmentation boundaries, the creping area is segmented to obtain a plurality of strip-shaped sub-areas; if the creping direction does not exist, a topological structure division sub-mechanism is executed to obtain a plurality of physically merged sub-areas; wherein the plurality of sub-areas include a plurality of physically merged sub-areas and a plurality of strip-shaped sub-areas.

3. A method of detecting and suppressing gravure printing wrinkles according to claim 2, wherein If the creping direction does not exist, a topological structure division sub-mechanism is executed to obtain a plurality of physically merged sub-areas, comprising: extracting the skeleton of the wrinkle in the pixel-level mask using a skeletonization algorithm, and identifying the branch points and end nodes of the skeleton line; taking the branch points and the end nodes as vertices, taking the skeleton path connecting the vertices as boundaries, and dividing the creping area into a plurality of topological polygon sub-areas; morphological optimization is performed on the plurality of topological polygon sub-areas, and adjacent sub-areas with an area less than a preset first threshold value and / or a shape irregularity greater than a preset second threshold value are merged to obtain a plurality of physically merged sub-areas.

4. A method of detecting and suppressing scudding in gravure printing according to claim 3, wherein Further comprising: calculating the number of sub-areas, the average area and the average shape irregularity of the plurality of physically merged sub-areas; comparing the number of sub-areas and the average shape irregularity with a preset target number range and a target average irregularity; if the number of sub-areas is greater than the upper limit of the target number range, the first threshold value is increased by a preset first step; if the number of sub-areas is less than the lower limit of the target number range, the first threshold value is reduced by the first step; if the average shape irregularity is greater than the target average irregularity, a preset second step is used to increase the second threshold value; an absolute value of a difference between the average area and a preset target average area is calculated to obtain an average area deviation; if the average area deviation is less than or equal to a preset third threshold value, the adjustment of the first threshold value and the second threshold value is continued until the number of sub-regions and the average shape irregularity meet the target number range and the target average irregularity at the same time; if the average area deviation is greater than the third threshold value, an adjustment process abnormality warning is triggered or the first threshold value is reset.

5. A method of detecting and suppressing gravure printing wrinkles according to claim 1, wherein the determination of the wrinkle suppression strategy based on the comprehensive influence level and the plurality of sub-regions comprises: determining a type of the wrinkle suppression strategy based on the plurality of sub-regions to obtain a preliminary suppression strategy; querying a preset influence level-basic adjustment amount mapping table based on the comprehensive influence level to obtain a corresponding basic adjustment amount of a printing machine unit corresponding to the wrinkle region; based on the basic adjustment amount, the preliminary suppression strategy, the distribution positions of the plurality of sub-regions and the sub-evaluation coefficient, spatial distribution weighting and priority allocation are performed on the basic adjustment amount to obtain the wrinkle suppression strategy.

6. A method of detecting and suppressing gravure printing wrinkles according to claim 5, wherein the determination of the type of the wrinkle suppression strategy based on the plurality of sub-regions to obtain a preliminary suppression strategy comprises: if the plurality of sub-regions are a plurality of belt-shaped sub-regions, the preliminary suppression strategy is determined to be a tension gradient adjustment strategy; if the plurality of sub-regions are a plurality of physically merged sub-regions, the preliminary suppression strategy is determined to be a unit cooperative adjustment strategy; if the wrinkle region has both belt-shaped sub-regions and physically merged sub-regions, the wrinkle region is determined to be a mixed type of wrinkle, and the preliminary suppression strategy is determined according to the projection distribution of all sub-regions in the paper running direction; if the belt-shaped sub-regions dominate in the projection distribution, the preliminary suppression strategy takes the tension gradient adjustment strategy as the dominant strategy and superimposes the unit cooperative adjustment strategy; if the physically merged sub-regions dominate in the projection distribution, the preliminary suppression strategy takes the unit cooperative adjustment strategy as the dominant strategy and superimposes the tension gradient adjustment strategy.

7. A method of detecting and suppressing gravure printing wrinkles according to claim 1, wherein before the generation of the suppression instruction based on the wrinkle suppression strategy and the adjustment of the printing machine unit corresponding to the wrinkle region based on the suppression instruction, the method further comprises: obtaining running state data of the printing machine unit corresponding to the wrinkle region; performing collaborative analysis and safety boundary verification on the parameter adjustment amount in the wrinkle suppression strategy and the running state data to obtain an analysis and verification result; optimizing and conflict resolving the parameter adjustment amount based on the analysis and verification result to obtain the wrinkle suppression strategy used for generating the suppression instruction.

8. An intaglio printing creping detection and suppression system characterized by, comprises: an image acquisition and processing module for acquiring linear array data of printed paper through a linear array camera and performing splicing to obtain image data; performing pre-processing and feature extraction on the image data to obtain feature image data; a wrinkle region determination module for inputting the feature image data into a pre-trained wrinkle evaluation model to obtain a wrinkle region and a pixel-level mask thereof; a wrinkled area division module configured to analyze a wrinkled direction and a wrinkle density of the wrinkled area according to the pixel-level mask, and divide the wrinkled area to obtain a plurality of sub-areas; an influence grade determination module configured to calculate optical properties and tension of all the sub-areas to obtain an optical abnormality index and a tension instability index, determine a sub-evaluation coefficient of each sub-area based on the optical abnormality index and the tension instability index, and determine a comprehensive influence grade of the wrinkled area according to a distribution feature of all the sub-evaluation coefficients; a suppression strategy determination module configured to determine a wrinkled suppression strategy based on the comprehensive influence grade and the plurality of sub-areas; an instruction generation and execution module configured to generate a suppression instruction based on the wrinkled suppression strategy, and adjust a printing machine unit corresponding to the wrinkled area based on the suppression instruction.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, The processor executes the computer program to implement the steps of the method of any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, the computer-readable storage medium comprising: The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 7.

Images