Method and system for automatically identifying and repairing defect data of image-text fast printing image

Abstract

The invention relates to the technical field of image-text fast printing, and discloses an image-text fast printing image defect data automatic identification and restoration method and system, and the method comprises the steps: obtaining optical scanning image data and digital design source data, so as to obtain an initial image set and a reference description set; according to object attributes, classifying pixels into different image-text areas, and according to the image-text areas, performing simulation rendering on the reference description set to obtain an expected printing image; performing multi-channel pixel-level comparison on the expected printing image and the initial image set to obtain a suspected defect area; analyzing the cause and defect type of the difference pixel set by combining the layer and path information to obtain a structured defect report; based on the structured defect report, performing targeted content reconstruction on the initial image set to obtain a repaired image; verifying the consistency of the restoration image and the reference description set to generate a restoration report; according to the invention, the accuracy of image-text fast printing image restoration can be improved.

Description

Technical Field

[0001] This invention relates to the field of rapid printing technology, and in particular to a method and system for automatic identification and repair of image defect data in rapid printing. Background Technology

[0002] In the field of image defect recognition in quick printing, existing technologies mostly rely on single-dimensional image comparison or simple pixel difference analysis, failing to fully integrate the correlation information between digital design source data and optical scan image data of quick printing products. In the process of pixel classification and image area division, there is a lack of precise correlation with page object attributes, resulting in a lack of targeted screening of difference pixels. It is difficult to effectively distinguish between process deviations and normal fluctuations in non-critical areas, leading to a high misjudgment rate of suspected defect areas. The completeness and accuracy of structured defect reports are insufficient, and they cannot provide reliable causal analysis and type definition basis for subsequent repairs.

[0003] In the image defect repair process, existing technologies often use a uniform repair algorithm to process defective areas without implementing targeted reconstruction based on defect type, image / text region category, and layer and path information in the baseline description set. This makes it difficult to balance the accuracy of contour reproduction and the consistency of color gradation during the repair process. Repaired images often exhibit problems such as blurred edges and color deviation from design standards. Furthermore, the naturalness of background transitions and overall coordination cannot be effectively guaranteed. Ultimately, this results in a low pass rate for consistency verification between the repaired image and the baseline description set, failing to meet the high-precision image quality requirements of quick printing. At the same time, the repair efficiency is low, making it difficult to adapt to the processing needs of batch quick printing products. Therefore, how to improve the accuracy of image repair in quick printing has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a method and system for automatic identification and repair of image defect data in quick printing, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an automatic identification and repair method for image defect data in quick printing, comprising:

[0006] S1. Acquire optical scanning image data and digital design source data of the graphic quick print to obtain the initial image set and reference description set of the graphic quick print;

[0007] S2. Based on the object attributes in the reference description set, the pixels in the initial image set are classified into different image and text regions, and the reference description set is simulated and rendered according to the image and text regions to obtain the expected printing image of the graphic quick print.

[0008] S3. Perform multi-channel pixel-level comparison between the expected printed image and the initial image set to obtain the suspected defect areas of the quick-printed graphic product;

[0009] S4. Combining the layer and path information in the benchmark description set, analyze the causes and defect types of the differential pixel sets in the suspected defect area to obtain a structured defect report of the graphic quick print.

[0010] S5. Based on the defect type and defect area category in the structured defect report, perform targeted content reconstruction on the initial image set to obtain the repaired image of the graphic print.

[0011] S6. Verify the consistency between the repaired image and the baseline description set to generate a repair report for the printed image.

[0012] In a preferred embodiment, acquiring optical scanned image data and digital design source data of the quick-printed graphic product to obtain an initial image set and a reference description set for the quick-printed graphic product includes:

[0013] Collect optical scan image data and corresponding digital design source data of quick-printed graphic materials;

[0014] Geometric correction is performed on the optical scan image data to obtain the standard image data of the quick-printed graphic product;

[0015] By integrating the page object description information, visual attribute information, and structural relationship information from the digital design source data, the original description dataset of the graphic quick print product is obtained;

[0016] In the spatial dimension, the standard image data and the original description dataset are aligned to obtain the initial image set and the baseline description set of the graphic print.

[0017] In a preferred embodiment, classifying pixels in the initial image set into different image regions based on object attributes in the baseline description set includes:

[0018] Extract the type identifier, geometric boundary description, and padding attributes of the page objects in the baseline description set;

[0019] Based on the geometric boundary description, the pixel region corresponding to the page object is identified from the initial image set;

[0020] Logically associate the color distribution features and texture complexity features of the pixel region with the type identifier and the fill attribute to obtain the association relationship between the pixel region and the page object;

[0021] Based on the aforementioned association, pixel regions with the same attribute descriptions are clustered to obtain the image and text regions of the initial image set.

[0022] In a preferred embodiment, the step of performing simulation rendering on the reference description set based on the image and text region to obtain the expected printed image of the quick-printed product includes:

[0023] The color and shape changes of the graphic area during the printing process are used as the characteristic data of the graphic area;

[0024] Based on the category of the image and text region and the characteristic data, visual simulation is performed on the original visual description of the page object in the benchmark description set to obtain the optimized visual description of the page object.

[0025] Encode the optimized visual description into rasterization rendering instructions for the page object;

[0026] The rasterization rendering instruction is applied to render the description data of the page object into a raster image of the page object;

[0027] By integrating the raster images, the expected printed image of the graphic quick print is obtained.

[0028] In a preferred embodiment, the step of performing a multi-channel pixel-level comparison between the expected printed image and the initial image set to obtain the suspected defect areas of the quick-printed image includes:

[0029] Multi-dimensional feature analysis is performed on the pixels corresponding to the expected printed image and the initial image set to obtain a multi-dimensional feature description of the pixels in the expected printed image.

[0030] The difference characteristics of the pixels are obtained by comparing the expected printed image with the multi-dimensional feature description.

[0031] The importance of the difference characteristics is evaluated based on the image / text region category to which the pixel belongs;

[0032] Based on the importance, the difference characteristics are fused to obtain a comprehensive difference index for the pixel;

[0033] The overall statistical distribution characteristics of the initial image set are used as the defect judgment criteria for the quick-printed graphic products;

[0034] Pixels whose comprehensive difference index exceeds the defect judgment benchmark are identified as initial abnormal pixels of the graphic prints;

[0035] Analyze the spatial connectivity of the initial abnormal pixels, and summarize the initial abnormal pixels that are spatially adjacent and have similar differences into the difference pixel set of the graphic quick print;

[0036] Based on the distribution contour of the difference pixel set, the corresponding closed region is defined in the initial image set to obtain the suspected defect region of the graphic print.

[0037] In a preferred embodiment, the step of fusing the difference characteristics based on the importance to obtain a comprehensive difference index for the pixel includes:

[0038] Tensor synthesis is performed on the importance and the difference characteristics to obtain the difference vector and weight vector of the pixel;

[0039] Based on the overall statistical distribution characteristics of the initial image set, multivariate correlation features between differences in different feature dimensions in the initial image set are extracted to construct the global covariance matrix of the initial image set;

[0040] Based on the difference vector, the weight vector, and the global covariance matrix, the comprehensive difference index of the pixel is calculated, wherein the formula for calculating the comprehensive difference index is:

[0041] ;

[0042] in, This indicates the comprehensive difference index. Represents the weight vector, This represents the difference vector. This indicates the preset temperature parameter. Denotes the global covariance matrix. This represents the preset adaptive adjustment vector. With vectors The elements are a diagonal matrix consisting of diagonal elements. This represents the preset regularization coefficient. This represents element-wise multiplication of vectors. Represents an exponential function. This represents the inversion of a vector.

[0043] In a preferred embodiment, the step of combining layer and path information from the baseline description set to analyze the causes and defect types of the differential pixel sets in the suspected defect area to obtain a structured defect report for the quick-printed graphic product includes:

[0044] Based on the difference pixel set of the suspected defect area, locate the corresponding design layer in the baseline description set, and parse the layer attributes and original path of the design layer;

[0045] The boundary contour of the difference pixel set is overlapped and compared with the path contour of the original path;

[0046] When the overlap of the overlap comparison is high, the initial cause of the difference pixel set is associated with the process deviation related to contour reproduction;

[0047] When the overlap of the overlap comparison is low, the fill attribute of the area covered by the difference pixel set in the design layer is extracted, and the consistency analysis of the fill attribute and the average visual attribute corresponding to the difference pixel set in the initial image set is performed.

[0048] If the results of the consistency analysis show that the average visual attribute deviates from the fill attribute, then the preliminary cause of the difference pixel set is associated with process deviations related to color and tone reproduction.

[0049] Based on the spatial location, preliminary cause, and corresponding design layer information of the differential pixel set, assign a corresponding defect type identifier to the differential pixel set;

[0050] Based on the defect type identifier and the detailed cause information of the difference pixel set, a structured defect report of the quick-printed graphic product is generated.

[0051] In a preferred embodiment, the step of reconstructing the initial image set based on the defect type and defect region category in the structured defect report to obtain the repaired image of the printed material includes:

[0052] Identify the defect type, image / text region category, and spatial extent of the defect region in the initial image set from the defect region in the structured defect report;

[0053] For defect types related to contour reconstruction, based on the path information of the corresponding design layer in the benchmark description set, the pixels within the defect range of the initial image set are reconstructed at the edges to obtain the contour reconstruction region of the initial image set.

[0054] For defect types related to color and tone reproduction, based on the fill attribute of the corresponding design layer in the benchmark description set and the normal pixel color distribution around the defect area in the initial image set, the pixels in the defect area are visually consistent and reconstructed to obtain the color reconstruction area of ​​the initial image set.

[0055] The restored image of the graphic print is generated by integrating the contour reconstruction region, the color reconstruction region, and the unmodified original image region in the initial image set.

[0056] In a preferred embodiment, verifying the consistency between the repaired image and the baseline description set to generate a repair report for the quick-printed image includes:

[0057] Spatially align the repaired image and the reference description set, and extract the repaired image feature description of the aligned repaired image and the reference feature description of the aligned reference description set;

[0058] The average color, texture complexity, and edge sharpness of the regions corresponding to the baseline feature description in the restored image feature description are summarized into the region visual features of the restored image feature description;

[0059] The visual attributes in the region visual features and the baseline feature description are compared item by item to obtain the visual deviation result of the repaired image feature description.

[0060] The background regions in the restored image that are not covered by the benchmark description set are identified, and the naturalness of the transition in color and texture and the overall coordination of the background regions are analyzed to obtain the continuity verification results of the restored image.

[0061] Based on the combined results of the visual deviation and the continuity verification, a repair report for the printed graphic product is generated.

[0062] To address the aforementioned problems, the present invention also provides an automatic identification and repair system for image defect data in rapid printing, the system comprising:

[0063] The data acquisition module is used to acquire optical scan image data and digital design source data of the graphic quick prints to obtain the initial image set and reference description set of the graphic quick prints.

[0064] The expected printed image simulation module is used to classify the pixels in the initial image set into different graphic regions according to the object attributes in the reference description set, and to perform simulation rendering on the reference description set according to the graphic regions to obtain the expected printed image of the graphic quick print.

[0065] The defect area identification module is used to perform multi-channel pixel-level comparison between the expected printed image and the initial image set to obtain the suspected defect areas of the quick-printed graphic product.

[0066] The defect report generation module is used to combine the layer and path information in the benchmark description set to analyze the cause and defect type of the difference pixel set in the suspected defect area, and obtain a structured defect report of the graphic quick print.

[0067] The image restoration module is used to perform targeted content reconstruction on the initial image set based on the defect type and defect area category in the structured defect report, so as to obtain the restored image of the graphic print.

[0068] The repair report generation module is used to verify the consistency between the repaired image and the baseline description set in order to generate a repair report for the printed image.

[0069] Compared with the prior art, the present invention has the following beneficial effects:

[0070] 1. This invention achieves pixel-level classification of image and text regions and multi-channel difference comparison by accurately integrating optical scan image data and digital design source data of quick-printed graphic materials. It also deeply analyzes the causes and types of defects by combining key information such as layers and paths in the benchmark description set, and constructs a structured defect report. This makes the defect identification process more targeted and comprehensive, greatly improves the accuracy and efficiency of image defect identification, and effectively avoids the omission or misjudgment of defects.

[0071] 2. This invention employs a targeted content reconstruction strategy based on defect type and region category to specifically repair defects related to contour reproduction, color, and tone. Simultaneously, through regional visual feature comparison and background region continuity verification, it ensures that the repaired image is highly consistent with the benchmark description set and has a harmonious and natural visual effect. This not only significantly improves the repair quality of quick-printed images but also simplifies the repair process, reduces invalid operations during the repair process, and further reduces the quality control costs and time consumption of quick-printed products. Attached Figure Description

[0072] Figure 1 This is a flowchart illustrating an automatic identification and repair method for image defect data in quick printing according to an embodiment of the present invention;

[0073] Figure 2 This is a functional module diagram of an automatic identification and repair system for image defect data in quick printing, provided in an embodiment of the present invention;

[0074] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0075] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0076] This application provides a method for automatic identification and repair of image defect data in quick-printing. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method for automatic identification and repair of image defect data in quick-printing can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0077] Reference Figure 1 The diagram shown is a flowchart illustrating an automatic identification and repair method for image defect data in quick printing according to an embodiment of the present invention. In this embodiment, the automatic identification and repair method for image defect data in quick printing includes:

[0078] S1. Acquire optical scanning image data and digital design source data of the graphic quick print to obtain the initial image set and reference description set of the graphic quick print;

[0079] In this embodiment of the invention, acquiring optical scan image data and digital design source data of the quick-printed graphic product to obtain an initial image set and a reference description set of the quick-printed graphic product includes:

[0080] Collect optical scan image data and corresponding digital design source data of quick-printed graphic materials;

[0081] Geometric correction is performed on the optical scan image data to obtain the standard image data of the quick-printed graphic product;

[0082] By integrating the page object description information, visual attribute information, and structural relationship information from the digital design source data, the original description dataset of the graphic quick print product is obtained;

[0083] In the spatial dimension, the standard image data and the original description dataset are aligned to obtain the initial image set and the baseline description set of the graphic print.

[0084] A high-precision optical scanner is used to perform a full-range scan of the graphic prints. During the scanning process, the scanner resolution is set to clearly capture all graphic details on the prints, including tiny text, pattern edges, and color transitions, thereby obtaining complete and detailed optical scan image data. At the same time, the original digital design source data corresponding to the prints is retrieved from the design storage terminal of the graphic print project. This ensures that the retrieved digital design source data and the scanned graphic prints belong to the same production project, guaranteeing an accurate correspondence between the two.

[0085] By analyzing the edge lines and key positioning points of the printed materials in the optical scan image, it is determined whether there are geometric deviations such as tilting, stretching, or local distortion in the image. For tilting issues, the image is rotated to make the edge parallel to the preset standard coordinate axis, using the horizontal or vertical edge of the printed material as a reference. For stretched or distorted parts, the distribution density of the image pixels is adjusted to make the proportion of each part in the image consistent with the actual physical size proportion of the printed materials. After the above operations, standard image data without geometric deviations and which can truly reflect the actual shape of the printed materials is obtained.

[0086] Each piece of information contained in the digital design source data is extracted one by one. The page object description information includes the specific type, shape features and size parameters of each page object. The visual attribute information includes the color value, brightness, contrast and texture style of each page object. The structural relationship information includes the hierarchical order of each page object on the page, the relative position coordinates and the connection or overlap relationship between them. These extracted page object description information, visual attribute information and structural relationship information are classified and organized according to the page object's affiliation to form an original description dataset covering all design details of quick print products.

[0087] Using the pixel coordinate system of standard image data as a unified reference framework, the specific corresponding position of each page object in the original description dataset in the coordinate system is determined. By adjusting the spatial reference parameters of the original description dataset, the position and size of each page object are made to completely coincide with the corresponding actual printed text and image in the standard image data, ensuring that the two are accurately matched in spatial position. After the spatial alignment operation, the original standard image data becomes the initial image set, and the original description dataset that has completed spatial alignment and has complete information becomes the benchmark description set.

[0088] The beneficial effects of this process are that, through precise acquisition, geometric correction, information integration, and spatial alignment, the initial image set can accurately reproduce the actual state of the printed materials, and the baseline description set can fully cover all the key information of the digital design. Moreover, the two are precisely corresponding in the spatial dimension, providing accurate and reliable basic data for subsequent classification of image and text areas based on the baseline description set, generation of expected printed images, and pixel-level comparison with the initial image set. This effectively avoids problems such as omissions and misjudgments in defect identification caused by data deviations or mismatches, laying a solid foundation for the efficient implementation of the entire automatic defect identification and repair process.

[0089] S2. Based on the object attributes in the reference description set, the pixels in the initial image set are classified into different image and text regions, and the reference description set is simulated and rendered according to the image and text regions to obtain the expected printing image of the graphic quick print.

[0090] In this embodiment of the invention, classifying pixels in the initial image set into different image and text regions based on object attributes in the reference description set includes:

[0091] Extract the type identifier, geometric boundary description, and padding attributes of the page objects in the baseline description set;

[0092] Based on the geometric boundary description, the pixel region corresponding to the page object is identified from the initial image set;

[0093] Logically associate the color distribution features and texture complexity features of the pixel region with the type identifier and the fill attribute to obtain the association relationship between the pixel region and the page object;

[0094] Based on the aforementioned association, pixel regions with the same attribute descriptions are clustered to obtain the image and text regions of the initial image set.

[0095] The step of performing simulation rendering on the reference description set based on the image and text region to obtain the expected printed image of the quick-printed product includes:

[0096] The color and shape changes of the graphic area during the printing process are used as the characteristic data of the graphic area;

[0097] Based on the category of the image and text region and the characteristic data, visual simulation is performed on the original visual description of the page object in the benchmark description set to obtain the optimized visual description of the page object.

[0098] Encode the optimized visual description into rasterization rendering instructions for the page object;

[0099] The rasterization rendering instruction is applied to render the description data of the page object into a raster image of the page object;

[0100] By integrating the raster images, the expected printed image of the graphic quick print is obtained.

[0101] The attribute information of each page object is retrieved one by one from the baseline description set. The type identifier must clearly distinguish the specific category of the page object, such as text object, graphic object, image object, etc., and each page object has a unique and clear type label. The geometric boundary description needs to extract the specific boundary parameters of the page object in the virtual coordinate system. For example, for a rectangular object, the coordinate values ​​of the upper left and lower right corners need to be obtained, and for a circular object, the coordinates of the center and the radius length need to be obtained, so as to accurately define the spatial range of the page object. The fill attribute needs to extract the details of the fill content of the page object, such as the specific color value of solid color fill, the color transition sequence of gradient fill, and the texture style and repetition frequency of texture fill, to ensure that the three core information of each page object—type identifier, geometric boundary description, and fill attribute—are fully obtained.

[0102] Based on the extracted geometric boundary description of the page object, pixel filtering is performed in the initial image set. If the geometric boundary of the page object is rectangular, all pixels whose coordinates fall within the coordinate range of the upper left and lower right corners of the rectangle are located in the initial image set. These pixels together constitute the pixel region corresponding to the page object. If the geometric boundary of the page object is circular, the distance from each pixel in the initial image set to the center coordinates of the circle is calculated, and all pixels whose distance is less than or equal to the radius are filtered out to form the pixel region corresponding to the circular page object, ensuring that each page object can find a precisely matching set of pixels in the initial image set.

[0103] First, feature analysis is performed on the identified pixel regions. Color distribution features are determined by statistically analyzing the color types of all pixels within the region and the percentage of each color. For example, if red pixels account for 95% and white pixels account for 5% in a certain pixel region, then the color distribution feature of that region is predominantly red and supplemented by white. Texture complexity features are judged by observing the frequency and pattern of pixel color changes within the region. If the pixel colors within the region are uniform and consistent with no obvious changes, the texture complexity is low. If the pixel colors within the region frequently alternate and there are complex patterns, the texture complexity is high. Then, these features are matched and associated with the type identifier and fill attribute of the page object. For example, for a page object with the type identifier "text" and the fill attribute "black solid color", its corresponding pixel region should exhibit the characteristics of concentrated color distribution and low texture complexity. If a pixel region meets this characteristic, the association between the pixel region and the page object is established.

[0104] Based on the established association between pixel regions and page objects, a unified standard for attribute descriptions is determined. Attribute descriptions consist of a combination of the page object's type identifier and fill attribute, such as "text - solid black" or "graphic - blue gradient." Then, all pixel regions with the same attribute description are selected. For example, all pixel regions associated with the attribute description "graphic - solid green" are summarized. The spatial position of these pixel regions is checked. If they are adjacent in the initial image set or belong to different parts of the same whole graphic, they are merged into a whole region. In this way, all pixel regions with different attribute descriptions are integrated, ultimately forming multiple clearly defined and attribute-unified graphic and text regions in the initial image set.

[0105] For each graphic area, we obtain its variation patterns during the printing process through actual printing tests. The color variation patterns need to record the deviation between the color of the graphic area and the color in the digital design after printing with specific inks. For example, the color of a certain yellow graphic area will shift to orange-yellow after printing, and the range of color values ​​after the shift needs to be clearly defined. The shape variation patterns need to record the changes of the edge of the graphic area during printing. For example, the edge of the text graphic area will have a blur range of 0.5 pixels due to ink diffusion. We systematically organize these color and shape variation patterns to form the unique characteristic data of each graphic area.

[0106] First, determine the category of the text and image area, such as "text-based text and image area" or "image-based text and image area." Then, based on the color variation patterns in the characteristic data, adjust the color of the original visual description of the corresponding page object in the baseline description set. For example, if the original visual description shows a certain text color as standard yellow, adjust the color to the orange-yellow that may appear after printing, according to the color variation patterns of that text-based text and image area. Based on the shape variation patterns in the characteristic data, adjust the edge details in the original visual description. For example, modify the original clear image edge description to an edge description with slight blurring. Through such visual simulation, obtain an optimized visual description of the page object that closely matches the actual printing effect.

[0107] The optimized visual description of the page object is decomposed into information, extracting key information such as color parameters, edge shape parameters, and size parameters. The color parameters are converted into specific RGB color value instructions, such as converting "orange-yellow" into the instruction format "RGB(255,165,0)". The edge shape parameters are converted into instructions for blur range and blur degree, such as converting "0.5 pixel blur" into the instruction "blur range: 0.5px, blur degree: light". The size parameters are converted into pixel size instructions, such as converting "100×200mm" into the instruction "1200×2400 pixels". All the decomposed and converted instructions are combined in a logical order to form the raster rendering instructions for the page object.

[0108] According to the pixel size instructions in the rasterization rendering instructions, the overall size of the raster image of the page object is determined; according to the RGB color value instructions, each pixel in the raster image is assigned a corresponding color; according to the edge blur instructions, the edge pixels of the raster image are processed with color transition to make the edges present a blur effect that meets the requirements of the instructions. Throughout the rendering process, all requirements of the rasterization rendering instructions are strictly followed to ensure that every detail is accurately presented, and finally the description data of the page object is transformed into a raster image composed of specific pixels.

[0109] Based on the structural relationship information of each page object in the baseline description set, determine the coordinate position of each page object's raster image within the overall page to ensure that the placement of the raster images is consistent with the design requirements. For raster images with hierarchical relationships, they are superimposed according to the hierarchical order in the baseline description set, with higher-level raster images covering lower-level raster images to avoid occlusion errors. All page object raster images are then combined according to the above rules to form an image that contains all graphic and text content and is highly consistent with the actual printing effect, i.e., the expected printing image of the graphic quick print.

[0110] The beneficial effects are that the above process accurately extracts page object attributes to achieve accurate classification of pixels in the initial image set, ensuring that the division of text and image areas is highly consistent with the design logic. At the same time, it constructs characteristic data based on actual printing rules and conducts simulation rendering, so that the expected printed image can realistically simulate the actual printing effect. This provides an accurate reference standard for subsequent multi-channel pixel-level comparison between the expected printed image and the initial image set, and avoids misjudgment of defects caused by large differences between the reference image and the actual printing. This lays a solid foundation for the accurate identification of image defects in quick printing.

[0111] S3. Perform multi-channel pixel-level comparison between the expected printed image and the initial image set to obtain the suspected defect areas of the quick-printed graphic product;

[0112] In this embodiment of the invention, the step of performing a multi-channel pixel-level comparison between the expected printed image and the initial image set to obtain the suspected defect areas of the quick-printed image includes:

[0113] Multi-dimensional feature analysis is performed on the pixels corresponding to the expected printed image and the initial image set to obtain a multi-dimensional feature description of the pixels in the expected printed image.

[0114] The difference characteristics of the pixels are obtained by comparing the expected printed image with the multi-dimensional feature description.

[0115] The importance of the difference characteristics is evaluated based on the image / text region category to which the pixel belongs;

[0116] Based on the importance, the difference characteristics are fused to obtain a comprehensive difference index for the pixel;

[0117] The overall statistical distribution characteristics of the initial image set are used as the defect judgment criteria for the quick-printed graphic products;

[0118] Pixels whose comprehensive difference index exceeds the defect judgment benchmark are identified as initial abnormal pixels of the graphic prints;

[0119] Analyze the spatial connectivity of the initial abnormal pixels, and summarize the initial abnormal pixels that are spatially adjacent and have similar differences into the difference pixel set of the graphic quick print;

[0120] Based on the distribution contour of the difference pixel set, the corresponding closed region is defined in the initial image set to obtain the suspected defect region of the graphic print.

[0121] The step of fusing the difference characteristics based on the importance to obtain the comprehensive difference index of the pixel includes:

[0122] Tensor synthesis is performed on the importance and the difference characteristics to obtain the difference vector and weight vector of the pixel;

[0123] Based on the overall statistical distribution characteristics of the initial image set, multivariate correlation features between differences in different feature dimensions in the initial image set are extracted to construct the global covariance matrix of the initial image set;

[0124] Based on the difference vector, the weight vector, and the global covariance matrix, the comprehensive difference index of the pixel is calculated, wherein the formula for calculating the comprehensive difference index is:

[0125] ;

[0126] in, This indicates the comprehensive difference index. Represents the weight vector, This represents the difference vector. This indicates the preset temperature parameter. Denotes the global covariance matrix. This represents the preset adaptive adjustment vector. With vectors The elements are a diagonal matrix consisting of diagonal elements. This represents the preset regularization coefficient. This represents element-wise multiplication of vectors. Represents an exponential function. This represents the inversion of a vector.

[0127] For each pixel whose coordinates completely correspond to those in the expected printed image and the initial image, the color feature, brightness feature, and texture feature of that pixel are extracted. The color feature is the specific value of the pixel in the three RGB channels, the brightness feature is the brightness value of the pixel calculated based on the RGB values, and the texture feature is a description of the color change pattern between the pixel and its eight neighboring pixels. The color feature value, brightness feature value, and texture feature description of the pixel in the expected printed image are arranged in a fixed order to form a multi-dimensional feature description of the pixel in the expected printed image.

[0128] The actual color features, actual brightness features, and actual texture features of the pixel in the expected printed image are taken and compared with the corresponding features of the pixel in the multi-dimensional feature description of the expected printed image. The numerical difference of each RGB channel in the color features and the numerical difference of the brightness features are calculated. It is also determined whether the change of the texture features exceeds the allowable range of the texture rules in the multi-dimensional feature description. These calculated differences and judgment results are organized according to feature type to form the difference characteristics of the pixel.

[0129] First, determine the text / image region category to which the pixel belongs. If the pixel belongs to the text region, since the text needs to be clearly distinguishable, differences in color and brightness features have a significant impact on text recognition. Therefore, the importance of these two types of differences is assessed as high, and the importance of texture feature differences is assessed as medium. If the pixel belongs to the graphic region, color feature differences directly affect the graphic appearance reproduction accuracy, so its importance is assessed as high, and the importance of brightness and texture feature differences is assessed as medium. If the pixel belongs to the background region, slight differences have little impact on the overall effect, so the importance of all differences is assessed as low. This completes the importance assessment of the difference feature.

[0130] Based on the importance of the assessed differences, corresponding weights are assigned to different types of differences. Differences with high importance are assigned higher weights, those with medium importance are assigned medium weights, and those with low importance are assigned lower weights. The value corresponding to each difference is multiplied by its weight, and all the product results are added together. The sum is the comprehensive difference index of that pixel.

[0131] The comprehensive difference index values ​​of all pixels in the initial image set are statistically analyzed, the frequency of these values ​​is recorded, the average value and dispersion of all values ​​are calculated, and the overall statistical distribution characteristics of the initial image set are determined based on these statistical results. The value at which the comprehensive difference index of 95% of the pixels in the overall statistical distribution characteristics does not exceed is taken as the boundary value, and this boundary value is set as the defect judgment benchmark for quick-printed graphic products.

[0132] Check the comprehensive difference index value of each pixel in the initial image set one by one, and compare the value with the defect judgment benchmark. If the comprehensive difference index value of a certain pixel is greater than the defect judgment benchmark value, it means that the difference between the pixel and the expected state exceeds the normal range, and the pixel is marked as the initial abnormal pixel of the graphic quick print.

[0133] For each initial anomalous pixel, check whether there are other initial anomalous pixels in its four adjacent positions (top, bottom, left, right) in the initial image set. If so, these pixels are determined to have spatial connectivity. Further compare the difference characteristics of these connected initial anomalous pixels to determine whether their difference types and degrees are similar. Group the initial anomalous pixels that are spatially connected and have similar difference characteristics into a group to form the difference pixel set of the graphic quick print.

[0134] Observe the distribution of the difference pixel set in the initial image set, find the outermost pixels in the set, and these outermost pixels constitute the distribution outline of the difference pixel set; connect these outermost pixels with continuous lines in a clockwise or counterclockwise order to form a complete closed line. The area enclosed by this closed line is the corresponding area defined in the initial image set, and this area is the suspected defect area of ​​the graphic print.

[0135] The pixel's difference characteristics are arranged in the order of color difference, brightness difference, and texture difference. The specific values ​​of each difference are arranged sequentially to form an ordered vector containing these values. This vector is the pixel's difference vector. At the same time, based on the importance evaluation results of the difference characteristics, the importance of color difference, brightness difference, and texture difference is converted into specific weight values ​​and arranged in the same order as the difference characteristics to form an ordered vector containing the weight values. This vector is the pixel's weight vector.

[0136] From the overall statistical distribution characteristics of the initial image set, the numerical distribution of color difference, brightness difference, and texture difference dimensions is extracted. The correlation between any two difference dimensions is calculated, such as whether the brightness difference increases as the color difference value increases. This correlation is the multivariate correlation feature between differences in different feature dimensions. The discreteness of each difference dimension and the correlation between different dimensions are arranged in a fixed row and column order, where the rows and columns correspond to the three dimensions of color difference, brightness difference, and texture difference. The values ​​at the intersection of the row and column are the covariance of the corresponding two dimensions, and the values ​​at the diagonal positions are the variance of the corresponding dimension. In this way, the global covariance matrix of the initial image set is constructed.

[0137] First, multiply each element of the weight vector with the corresponding element of the difference vector to obtain a new intermediate vector. Then, perform the inverse operation on the global covariance matrix to obtain the global covariance inverse matrix. Next, multiply the intermediate vector with the global covariance inverse matrix to obtain another intermediate result vector. Then, multiply this intermediate result vector with the vector after transposing the intermediate vector to obtain a single value. Finally, perform the square root operation on this single value, and the result is the comprehensive difference index of the pixel.

[0138] The importance of the difference characteristics is assessed based on the image and text region category to which the pixel belongs. This importance is then combined with the difference characteristics using tensor synthesis. The process involves first identifying the image and text region category to which the pixel belongs, determining the importance of the difference characteristics in the overall judgment based on this category, and then combining this importance with the difference characteristics obtained by comparing the expected printed image and multi-dimensional feature descriptions using tensor synthesis to finally form a weight vector.

[0139] The difference vector is generated through the tensor synthesis process of importance and difference characteristics mentioned above. The difference characteristics are obtained by first performing multi-dimensional feature analysis on the corresponding pixels in the expected printed image and the initial image set to obtain a multi-dimensional feature description of the pixels in the expected printed image, and then comparing the expected printed image with the multi-dimensional feature description. The difference vector is obtained by tensor synthesis of this result and importance.

[0140] The temperature parameter is a fixed value that is preset before the calculation begins and is used for subsequent exponential function calculations and adjustments.

[0141] The process of constructing the global covariance matrix involves first obtaining the overall statistical distribution characteristics of the initial image set, then extracting the multivariate correlation features between the differences in different feature dimensions from these characteristics, and finally calculating the global covariance matrix according to the construction rules of the covariance matrix based on these multivariate correlation features.

[0142] The adaptive adjustment vector is a pre-defined vector before calculation, with each element corresponding to an adjustment coefficient for a different feature dimension, which is used for subsequent adaptive adjustment of the global covariance matrix.

[0143] The regularization coefficient is a fixed value set before calculation to avoid singular values ​​in matrix operations and ensure the stability of the calculation.

[0144] By performing element-wise multiplication of the weight vector and the difference vector processed by the exponential function to obtain the intermediate vector, and then inverting the intermediate vector, it is multiplied by the inverse square root matrix of the global covariance matrix adjusted by the regularization coefficient and the adaptive adjustment vector. Finally, the result is multiplied by the previous intermediate vector and the square root is taken. The whole process comprehensively considers the importance of pixel difference characteristics, the correlation of differences in different feature dimensions, the stability of the operation, and the need for adaptive adjustment. The final value can accurately quantify the comprehensive difference between the pixel and the expected printed image corresponding to the benchmark description set. This value is used to determine whether the pixel belongs to the initial abnormal pixel and provides the core basis for the subsequent identification of suspected defect areas.

[0145] The more significant the difference characteristics of pixels, that is, the more obvious the difference in multi-dimensional features between the corresponding pixels in the expected printed image and the initial image set, the larger the value of the difference vector after the exponential function processing will be. The value of the intermediate vector obtained after element-wise multiplication with the weight vector will also increase, which in turn increases the overall difference value after the multiplication operation and the square root operation.

[0146] The higher the importance of the image region category to which a pixel belongs, the larger the value of the corresponding element in the weight vector. When performing element-wise multiplication with the difference vector after exponential function processing, the value of the intermediate vector will increase accordingly, ultimately leading to an increase in the overall difference value.

[0147] The stronger the correlation between differences in different feature dimensions, the corresponding values ​​of the elements in the global covariance matrix will change accordingly. After adding the diagonal matrix formed by the regularization coefficient and the adaptive adjustment vector, the inverse square root matrix is ​​then calculated. When this matrix is ​​multiplied by the intermediate vector and its inverse vector, the overall difference value can better reflect the true difference under the overall data distribution, avoiding the excessive amplification or reduction of differences in a single dimension.

[0148] The regularization coefficient is added to the global covariance matrix by the diagonal matrix formed by the adaptive adjustment vector, which adjusts the value of the global covariance matrix. This makes the operation of finding the square root of the matrix inverse more stable, and avoids situations where there is no solution or abnormal values. As a result, the change of the overall difference value is more stable, and extreme values ​​will not occur due to data fluctuations.

[0149] Each element of the adaptive adjustment vector corresponds to a different feature dimension. When a certain feature dimension has a greater impact on the difference judgment in a specific image and text area, the value of the corresponding element in the vector will make the weight of the adjusted global covariance matrix on that feature dimension higher, thereby making the comprehensive difference value more in line with the characteristics of the image and text area. When the difference of that feature dimension increases, the comprehensive difference value will change more sensitively.

[0150] The beneficial effects are as follows: the above process achieves accurate comparison between the expected and initial images through multi-dimensional feature analysis, and combines the evaluation of the importance of difference characteristics with the category of image and text regions, making the calculation of comprehensive difference index more targeted and avoiding interference from irrelevant differences; the defect judgment benchmark is determined based on the overall statistical distribution of the initial image set, ensuring that the benchmark conforms to the overall characteristics of the image and reducing subjective judgment errors; through spatial connectivity analysis, the difference pixel set is summarized and closed regions are defined, accurately locating suspected defect areas, and the comprehensive difference index calculation incorporates feature dimension correlation, further improving the reliability of judgment, effectively reducing the probability of false and false defects, providing accurate targets for subsequent defect cause analysis and repair, and significantly improving the accuracy and efficiency of image defect recognition in quick printing.

[0151] S4. Combining the layer and path information in the benchmark description set, analyze the causes and defect types of the differential pixel sets in the suspected defect area to obtain a structured defect report of the graphic quick print.

[0152] In this embodiment of the invention, the step of combining the layer and path information in the benchmark description set to analyze the causes and defect types of the differential pixel sets in the suspected defect area, and obtaining a structured defect report of the printed material, includes:

[0153] Based on the difference pixel set of the suspected defect area, locate the corresponding design layer in the baseline description set, and parse the layer attributes and original path of the design layer;

[0154] The boundary contour of the difference pixel set is overlapped and compared with the path contour of the original path;

[0155] When the overlap of the overlap comparison is high, the initial cause of the difference pixel set is associated with the process deviation related to contour reproduction;

[0156] When the overlap of the overlap comparison is low, the fill attribute of the area covered by the difference pixel set in the design layer is extracted, and the consistency analysis of the fill attribute and the average visual attribute corresponding to the difference pixel set in the initial image set is performed.

[0157] If the results of the consistency analysis show that the average visual attribute deviates from the fill attribute, then the preliminary cause of the difference pixel set is associated with process deviations related to color and tone reproduction.

[0158] Based on the spatial location, preliminary cause, and corresponding design layer information of the differential pixel set, assign a corresponding defect type identifier to the differential pixel set;

[0159] Based on the defect type identifier and the detailed cause information of the difference pixel set, a structured defect report of the quick-printed graphic product is generated.

[0160] Based on the spatial coordinate range of the differential pixel set in the suspected defect area, a search is performed in the layer list of the baseline description set to find all design layers whose spatial range intersects with the coordinate range. The design layer most closely associated with the text and image content corresponding to the differential pixel set is then identified. For example, if the differential pixel set corresponds to text content, the design layer containing the text is located. Subsequently, the layer attributes of this design layer are parsed, including the layer name, its hierarchical order in the overall page, whether transparency is set, and the transparency parameters. At the same time, the original path of the page object in this design layer is parsed. If the original path is a vector path, the node coordinates, the line segment type between nodes, and the control point coordinates of the curve are extracted. If it is a regular graphic path such as a rectangle or circle, the geometric parameters of the graphic are extracted, such as the length, width, and top-left corner coordinates of the rectangle, and the center coordinates and radius of the circle, thereby obtaining the corresponding design layer, layer attributes, and original path.

[0161] First, extract the boundary contour of the difference pixel set. By traversing all pixels in the difference pixel set, select pixels that are only adjacent to non-difference pixels as boundary pixels. Record the coordinates of these boundary pixels in clockwise order and connect them to form the boundary contour of the difference pixel set. Then, based on the original path parameters obtained from the parsing, draw the contour of the original path in the virtual coordinate system and record the coordinates of all points on the contour. Next, calculate the number of coordinate points in the overlapping part of the two contours. Divide the number of overlapping coordinate points by the total number of coordinate points in the boundary contour of the difference pixel set to obtain the overlap value. This completes the overlap comparison between the boundary contour of the difference pixel set and the path contour of the original path.

[0162] When the calculated overlap value is greater than or equal to the preset high overlap threshold, it indicates that the boundary of the difference pixel set basically coincides with the original path outline. The difference is mainly reflected in the integrity or accuracy of the outline. At this time, the preliminary cause of the difference pixel set is associated with the process deviation related to outline reproduction, specifically including outline offset caused by misregistration during printing, blurring of outline edges caused by plate wear, and local missing outline caused by cutting position deviation, thus clarifying the specific type of the preliminary cause.

[0163] When the overlap value is less than the preset high overlap threshold, the fill attribute is extracted from the located design layer attributes. If the design layer is a solid color fill, the RGB or CMYK color value of the fill color is extracted; if it is a gradient fill, the gradient start color, end color, and gradient direction parameters are extracted; if it is a texture fill, the original file of the texture pattern and the repeating arrangement parameters are extracted. At the same time, the visual attributes of all pixels in the difference pixel set are statistically analyzed. The values ​​of each color channel are summed and divided by the total number of pixels to obtain the average visual attribute of the difference pixel set in the initial image set. By comparing the differences between the average visual attribute and the fill attribute, such as the difference in color channel values ​​and the matching degree of texture style, the consistency analysis is completed.

[0164] If the consistency analysis results show that the difference between the average visual attribute and the fill attribute exceeds the normal printing error range, such as the difference between any channel in the RGB color value exceeding 20, or the texture pattern and the texture pattern in the fill attribute being completely mismatched, that is, the average visual attribute deviates from the fill attribute, then the preliminary cause of the difference pixel set is associated with the process deviation related to color and tone reproduction. Specifically, this includes color distortion caused by ink mixing ratio deviation, loss of tone levels caused by uneven printing pressure, color fading caused by improper drying temperature, and other process problems, to determine the specific content of the preliminary cause.

[0165] Obtain the spatial location information of the difference pixel set, that is, the coordinate range formed by the minimum and maximum coordinates of the difference pixel set in the initial image set; sort out the preliminary causes corresponding to the difference pixel set, such as "misregistration leading to outline offset" and "ink ratio deviation leading to color distortion" and the corresponding design layer information; according to the preset defect type classification rules, assign a defect type label to the difference pixel set. For example, if the preliminary cause is related to the process deviation of outline reproduction and corresponds to the text layer, it is labeled as "text outline offset defect". If the preliminary cause is related to the process deviation of color and tone reproduction and corresponds to the graphic layer, it is labeled as "graphic color distortion defect".

[0166] The information corresponding to each set of differing pixels is organized into a structured defect report according to a unified structure. The report includes fields such as "defect number", "corresponding design layer information", "defect spatial location", "preliminary cause of defect", and "defect type identifier". The information of all sets of differing pixels is classified and summarized according to the defect type identifier. Under each defect type, the corresponding entries are arranged in order by defect number. At the same time, defect statistics are added at the end of the report to form a complete structured defect report for graphic quick print products.

[0167] The beneficial effects are that this process, by combining the layer and path information of the baseline description set, achieves accurate positioning of the causes of differential pixel integration, avoiding the blindness of judging the cause solely based on the appearance of the image; the assignment of defect type identifiers makes defect classification clear and specific, and the structured defect report systematically integrates the key information of the defect, providing detailed and accurate basis for subsequent targeted repairs, effectively reducing the trial and error costs in the repair process, and improving the efficiency and accuracy of image defect repair in quick printing.

[0168] S5. Based on the defect type and defect area category in the structured defect report, perform targeted content reconstruction on the initial image set to obtain the repaired image of the graphic print.

[0169] In this embodiment of the invention, the step of performing targeted content reconstruction on the initial image set based on the defect type and defect region category in the structured defect report to obtain the repaired image of the printed material includes:

[0170] Identify the defect type, image / text region category, and spatial extent of the defect region in the initial image set from the defect region in the structured defect report;

[0171] For defect types related to contour reconstruction, based on the path information of the corresponding design layer in the benchmark description set, the pixels within the defect range of the initial image set are reconstructed at the edges to obtain the contour reconstruction region of the initial image set.

[0172] For defect types related to color and tone reproduction, based on the fill attribute of the corresponding design layer in the benchmark description set and the normal pixel color distribution around the defect area in the initial image set, the pixels in the defect area are visually consistent and reconstructed to obtain the color reconstruction area of ​​the initial image set.

[0173] The restored image of the graphic print is generated by integrating the contour reconstruction region, the color reconstruction region, and the unmodified original image region in the initial image set.

[0174] Retrieve detailed entries for each defect area from the structured defect report, and examine the "Defect Type Identifier" field to determine if the defect area belongs to a defect type related to contour reproduction, such as "text contour offset defect" or "graphic contour blurring defect," or a defect type related to color and tone reproduction, such as "solid color fill color distortion defect" or "gradient fill tone loss defect." Link the defect area to the baseline description set through the "corresponding design layer information" in the report, and determine the graphic / text area category (text area, graphic area, background area) based on the page object attributes corresponding to the design layer. Read the coordinate range recorded in the "Defect Spatial Location" field of the report, such as the coordinates from the upper left corner (x1, y1) to the lower right corner (x2, y2), to accurately locate the spatial range of the defect area in the initial image set, and fully obtain the defect type, graphic / text area category, and spatial range information for each defect area.

[0175] For defect types related to contour reconstruction, path information of the corresponding design layer is extracted from the baseline description set. If the path is a regular graphic path such as a rectangle or circle, the geometric parameters of the graphic are obtained, such as the length, width, and vertex coordinates of the rectangle, and the center coordinates and radius of the circle. If the path is an irregular vector path, the coordinates of all nodes on the path, the type of line segments between nodes, and the coordinates of the control points of the curve are extracted. The spatial range of the defect area is located in the initial image set. Using the extracted path information as a standard template, the pixel edges within the defect range are adjusted: for defects with contour offset, the edge pixels that deviate from the path are moved to the coordinate positions specified by the path, so that the edges fit the path perfectly; for defects with blurred contours, the color contrast between edge pixels and adjacent non-edge pixels is enhanced according to the clear edge boundaries of the path, and the transition colors of blurred pixels are eliminated, so that the edge lines are clear and sharp. After the above edge reconstruction operation, an area that meets the requirements of the baseline path is formed within the defect range of the initial image set, which is the contour reconstruction area.

[0176] For defect types related to color and tone reproduction, firstly, obtain the fill attributes of the corresponding design layer from the baseline description set: if it is a solid color fill, extract the standard RGB or CMYK color values ​​of the fill color; if it is a gradient fill, extract the color values ​​of the gradient start color and end color, as well as the gradient direction and gradient transition length parameters; if it is a texture fill, extract the original pattern file of the texture and the repeating density parameters of the pattern in the layer. Then, find normal pixels within a 10-20 pixel range around the defect area in the initial image set—that is, pixels that are not marked as defects and meet design expectations—and statistically analyze the color distribution characteristics of these normal pixels, including the color brightness gradient, the color value transition difference between adjacent pixels, and the overall hue. The process involves adjusting the color and tone of each pixel within the defective area, based on the standard color determined by the fill attribute and combined with the color distribution characteristics of surrounding normal pixels. For solid color-filled defective areas, the color value of each pixel is corrected to the standard color value, and the pixel brightness is fine-tuned to ensure a natural transition with the surrounding area, referencing the brightness of surrounding normal pixels. For gradient-filled defective areas, pixel color values ​​are redistributed according to gradient parameters to ensure that the color value changes conform to the gradient direction and transition length, while matching the hue of surrounding normal pixels. For texture-filled defective areas, texture pixels within the defective area are regenerated according to the standard texture pattern to ensure the continuity and clarity of the texture. The adjusted defective area is the color reconstruction area.

[0177] Using the pixel coordinate system of the initial image set as a unified benchmark, the contour reconstruction region and color reconstruction region are superimposed onto their respective spatial ranges in the initial image set, ensuring that the coordinates of each pixel in the reconstruction region completely correspond to the coordinates of the initial image set without offset or misalignment. The connections between the contour reconstruction region and the surrounding unmodified original image region, the color reconstruction region and the surrounding unmodified original image region, and the contour reconstruction region and the color reconstruction region are checked. If there are obvious abrupt differences in color, brightness, or edges of the pixels at the connection point, pixels within a range of 5-8 pixels at the connection point are selected for fine-tuning to gradually transition the attributes of the pixels at the connection point and eliminate the sense of difference. The adjusted contour reconstruction region and color reconstruction region are integrated with the unmodified original image region in the initial image set to form a complete, coherent image that meets the design requirements of the benchmark description set. This image is the restored image of the graphic design product.

[0178] The beneficial effects of this process are as follows: First, it accurately identifies key defect information, enabling targeted reconstruction of different types of defects and avoiding image distortion caused by indiscriminate repair; second, the contour reconstruction strictly follows the baseline path information, ensuring the accuracy and standardization of image and text edges and solving problems such as contour offset and blurring; third, the color reconstruction combines fill attributes with the distribution of surrounding normal pixels, ensuring the consistency of the color of the defect area with the design standard and achieving a natural transition with the surrounding image, avoiding repair traces; finally, the region integration operation further optimizes the overall image coordination, enabling the generated repaired image to highly restore the expected effect of the digital design source data, effectively improving the accuracy and overall quality of image defect repair in quick printing, while the operation process is clear and controllable, and can be efficiently applied to defect repair scenarios of various quick printing products.

[0179] S6. Verify the consistency between the repaired image and the baseline description set to generate a repair report for the printed image.

[0180] In this embodiment of the invention, verifying the consistency between the repaired image and the baseline description set to generate a repair report for the quick-printed image includes:

[0181] Spatially align the repaired image and the reference description set, and extract the repaired image feature description of the aligned repaired image and the reference feature description of the aligned reference description set;

[0182] The average color, texture complexity, and edge sharpness of the regions corresponding to the baseline feature description in the restored image feature description are summarized into the region visual features of the restored image feature description;

[0183] The visual attributes in the region visual features and the baseline feature description are compared item by item to obtain the visual deviation result of the repaired image feature description.

[0184] The background regions in the restored image that are not covered by the benchmark description set are identified, and the naturalness of the transition in color and texture and the overall coordination of the background regions are analyzed to obtain the continuity verification results of the restored image.

[0185] Based on the combined results of the visual deviation and the continuity verification, a repair report for the printed graphic product is generated.

[0186] Using the boundary coordinates and center point coordinates of each page object in the baseline description set as reference standards, the pixel spatial positions of the restored image are adjusted so that the boundary coordinates and center point coordinates of each page object in the restored image completely coincide with the coordinates of the corresponding page object in the baseline description set, thus achieving spatial alignment between the restored image and the baseline description set. After alignment, for each region in the restored image that corresponds one-to-one with the page objects in the baseline description set, the average RGB color channel values ​​of all pixels in that region are calculated as the region's average color. The ratio of the number of significant color changes in pixels within that region to the total number of pixels in that region is used as the texture complexity. The brightness difference between edge pixels and their directly adjacent non-edge pixels is measured as the edge sharpness. These data are categorized and organized according to their corresponding regions to form a feature description of the restored image. Simultaneously, the fill attributes of each page object are extracted from the baseline description set, such as the standard RGB color values ​​for solid color fills, the color value range for gradient fills, the texture density parameters for texture fills, and texture description information, such as the repetition frequency of texture patterns and texture detail features. Path edge specifications, such as the required brightness difference standard for edges, are also included. These are categorized and organized according to page objects to form a baseline feature description.

[0187] For each region in the restored image feature description that corresponds to the baseline feature description, the three data points of the region's average color, texture complexity, and edge sharpness are extracted one by one and arranged in an orderly manner according to the format of region identifier - average color - texture complexity - edge sharpness to ensure that the three feature data points of each corresponding region are fully presented. The data of all corresponding regions are summarized and integrated to form the regional visual features of the restored image feature description.

[0188] The color comparison is performed as follows: First, the average color of the region in the visual features is compared with the fill color value of the corresponding page object in the baseline feature description. The difference between the values ​​of each RGB channel is calculated, and it is determined whether the difference is within the preset normal error range (i.e., the difference between each channel does not exceed 5). The color comparison result is recorded. Second, the texture complexity of the region in the visual features is compared with the texture description parameters of the corresponding page object in the baseline feature description, such as the color change frequency standard corresponding to the texture repetition frequency. The texture complexity of the region in the restored image is determined to be consistent with the baseline requirements, and the texture comparison result is recorded. Third, the edge sharpness of the region in the visual features is compared with the path edge specifications of the corresponding page object in the baseline feature description, such as the minimum value that the edge brightness difference should reach. The edge sharpness of the region in the restored image is determined to be consistent with the baseline requirements, and the edge comparison result is recorded. Finally, the comparison results of color, texture, and edge are organized by region to clarify whether there is a deviation, the type of deviation, and the degree of deviation in each region, forming the visual deviation result of the feature description of the restored image.

[0189] Traverse all pixel coordinates of the restored image and compare them with the boundary coordinate ranges of all page objects in the baseline description set. Filter out pixels that do not fall within the boundary coordinate range of any page object. The continuous area formed by these pixels is the background area in the restored image that is not covered by the baseline description set. Observe the color transition between the background area and the adjacent image and text areas, that is, the areas covered by the baseline description set, and determine whether the color changes gradually from the edge of the image and text area to the background area without obvious color block breaks, so as to analyze the naturalness of the color transition. Check the color distribution and texture style inside the background area to determine whether there are local color abrupt changes or texture inconsistencies, so as to analyze the overall coordination. Organize the analysis results of the naturalness of the color transition and the overall coordination to form the continuity verification results of the restored image.

[0190] The details of deviations in each region of the visual deviation results, such as deviation region identification, deviation type, deviation degree, and continuity verification results, such as the evaluation of the naturalness of background area transition and the overall coordination evaluation, are integrated and organized according to the structure of repair verification overview - visual deviation details - background continuity analysis - verification conclusion. The repair verification overview briefly describes the verification object, namely the repaired image, the benchmark description set, and the verification process. The visual deviation details list the comparison results for each region. The background continuity analysis presents the evaluation results of the background area. The verification conclusion clarifies whether the repaired image meets the requirements of the benchmark description set. For example, if there is no deviation and the background is coordinated, it is judged as qualified. If there is obvious deviation, it is judged as needing further optimization, thus forming a repair report for quick-printed graphics.

[0191] The beneficial effects of this process are as follows: spatial alignment ensures accurate comparison between the restored image and the baseline description set, avoiding verification errors caused by positional deviations; item-by-item comparison of regional visual features can comprehensively identify differences between the restored image and the design baseline, ensuring the restoration accuracy of core features of the image and text; the continuity analysis of the background area compensates for the blind spot of focusing only on the design object, ensuring the overall coordination of the image effect; the final generated restoration report clearly presents the verification results and conclusions, providing a clear basis for judging the restoration effect and pointing the way for possible secondary restorations, effectively improving the comprehensiveness, accuracy and practicality of image restoration verification in quick printing.

[0192] like Figure 2 The diagram shown is a functional module diagram of an automatic identification and repair system for image defect data in quick printing provided by an embodiment of the present invention.

[0193] The automatic identification and repair system 100 for image defect data in rapid printing, as described in this invention, can be installed in an electronic device. Depending on the functions implemented, the automatic identification and repair system 100 for image defect data in rapid printing may include a data acquisition module 101, a expected printed image simulation module 102, a defect area identification module 103, a defect report generation module 104, an image repair module 105, and a repair report generation module 106. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, and are stored in the memory of the electronic device.

[0194] In this embodiment, the functions of each module / unit are as follows:

[0195] The data acquisition module 101 is used to acquire optical scan image data and digital design source data of the graphic quick print, so as to obtain the initial image set and reference description set of the graphic quick print;

[0196] The expected printed image simulation module 102 is used to classify the pixels in the initial image set into different graphic regions according to the object attributes in the reference description set, and to perform simulation rendering on the reference description set according to the graphic regions to obtain the expected printed image of the graphic quick print.

[0197] The defect area identification module 103 is used to perform multi-channel pixel-level comparison between the expected printed image and the initial image set to obtain the suspected defect areas of the graphic quick print.

[0198] The defect report generation module 104 is used to combine the layer and path information in the benchmark description set to analyze the cause and defect type of the difference pixel set in the suspected defect area, and obtain a structured defect report of the graphic quick print.

[0199] The image restoration module 105 is used to perform targeted content reconstruction on the initial image set based on the defect type and defect area category in the structured defect report, so as to obtain the restored image of the graphic print.

[0200] The repair report generation module 106 is used to verify the consistency between the repaired image and the baseline description set in order to generate a repair report for the printed graphic product.

[0201] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0202] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0203] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0204] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0205] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0206] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for automatic identification and repair of image defect data in quick printing, characterized in that, The method includes: S1. Acquire optical scanning image data and digital design source data of the graphic quick print to obtain the initial image set and reference description set of the graphic quick print; S2. Based on the object attributes in the reference description set, the pixels in the initial image set are classified into different image and text regions, and the reference description set is simulated and rendered according to the image and text regions to obtain the expected printing image of the graphic quick print. S3. Perform multi-channel pixel-level comparison between the expected printed image and the initial image set to obtain the suspected defect areas of the quick-printed graphic product; S4. Combining the layer and path information in the benchmark description set, analyze the causes and defect types of the differential pixel sets in the suspected defect area to obtain a structured defect report of the graphic quick print. S5. Based on the defect type and defect area category in the structured defect report, perform targeted content reconstruction on the initial image set to obtain the repaired image of the graphic print. S6. Verify the consistency between the repaired image and the baseline description set to generate a repair report for the printed image.

2. The method for automatic identification and repair of image defect data in quick printing as described in claim 1, characterized in that, The process of acquiring optical scanned image data and digital design source data of the quick-printed graphic materials to obtain an initial image set and a reference description set for the quick-printed graphic materials includes: Collect optical scan image data and corresponding digital design source data of quick-printed graphic materials; Geometric correction is performed on the optical scan image data to obtain the standard image data of the quick-printed graphic product; By integrating the page object description information, visual attribute information, and structural relationship information from the digital design source data, the original description dataset of the graphic quick print product is obtained; In the spatial dimension, the standard image data and the original description dataset are aligned to obtain the initial image set and the baseline description set of the graphic print.

3. The method for automatic identification and repair of image defect data in quick printing as described in claim 1, characterized in that, The step of classifying pixels in the initial image set into different image regions based on object attributes in the baseline description set includes: Extract the type identifier, geometric boundary description, and padding attributes of the page objects in the baseline description set; Based on the geometric boundary description, the pixel region corresponding to the page object is identified from the initial image set; Logically associate the color distribution features and texture complexity features of the pixel region with the type identifier and the fill attribute to obtain the association relationship between the pixel region and the page object; Based on the aforementioned association, pixel regions with the same attribute descriptions are clustered to obtain the image and text regions of the initial image set.

4. The method for automatic identification and repair of image defect data in quick printing as described in claim 1, characterized in that, The step of performing simulation rendering on the reference description set based on the image and text region to obtain the expected printed image of the quick-printed product includes: The color and shape changes of the graphic area during the printing process are used as the characteristic data of the graphic area; Based on the category of the image and text region and the characteristic data, visual simulation is performed on the original visual description of the page object in the benchmark description set to obtain the optimized visual description of the page object. Encode the optimized visual description into rasterization rendering instructions for the page object; The rasterization rendering instruction is applied to render the description data of the page object into a raster image of the page object; By integrating the raster images, the expected printed image of the graphic quick print is obtained.

5. The method for automatic identification and repair of image defect data in quick printing as described in claim 1, characterized in that, The step of performing a multi-channel pixel-level comparison between the expected printed image and the initial image set to obtain the suspected defect areas of the quick-printed image includes: Multi-dimensional feature analysis is performed on the pixels corresponding to the expected printed image and the initial image set to obtain a multi-dimensional feature description of the pixels in the expected printed image. The difference characteristics of the pixels are obtained by comparing the expected printed image with the multi-dimensional feature description. The importance of the difference characteristics is evaluated based on the image / text region category to which the pixel belongs; Based on the importance, the difference characteristics are fused to obtain a comprehensive difference index for the pixel; The overall statistical distribution characteristics of the initial image set are used as the defect judgment criteria for the quick-printed graphic products; Pixels whose comprehensive difference index exceeds the defect judgment benchmark are identified as initial abnormal pixels of the graphic prints; Analyze the spatial connectivity of the initial abnormal pixels, and summarize the initial abnormal pixels that are spatially adjacent and have similar differences into the difference pixel set of the graphic quick print; Based on the distribution contour of the difference pixel set, the corresponding closed region is defined in the initial image set to obtain the suspected defect region of the graphic print.

6. The method for automatic identification and repair of image defect data in quick printing as described in claim 5, characterized in that, The step of fusing the difference characteristics based on the importance to obtain the comprehensive difference index of the pixel includes: Tensor synthesis is performed on the importance and the difference characteristics to obtain the difference vector and weight vector of the pixel; Based on the overall statistical distribution characteristics of the initial image set, multivariate correlation features between differences in different feature dimensions in the initial image set are extracted to construct the global covariance matrix of the initial image set; Based on the difference vector, the weight vector, and the global covariance matrix, the comprehensive difference index of the pixel is calculated, wherein the formula for calculating the comprehensive difference index is: ； in, This indicates the comprehensive difference index. Represents the weight vector, This represents the difference vector. This indicates the preset temperature parameter. Denotes the global covariance matrix. This represents the preset adaptive adjustment vector. With vectors The elements are a diagonal matrix consisting of diagonal elements. This represents the preset regularization coefficient. This represents element-wise multiplication of vectors. Represents an exponential function. This represents the inversion of a vector.

7. The method for automatic identification and repair of image defect data in quick printing as described in claim 1, characterized in that, The process involves combining layer and path information from the baseline description set to analyze the causes and defect types of the differential pixel sets in the suspected defect areas, resulting in a structured defect report for the printed material, including: Based on the difference pixel set of the suspected defect area, locate the corresponding design layer in the baseline description set, and parse the layer attributes and original path of the design layer; The boundary contour of the difference pixel set is overlapped and compared with the path contour of the original path; When the overlap of the overlap comparison is high, the initial cause of the difference pixel set is associated with the process deviation related to contour reproduction; When the overlap of the overlap comparison is low, the fill attribute of the area covered by the difference pixel set in the design layer is extracted, and the consistency analysis of the fill attribute and the average visual attribute corresponding to the difference pixel set in the initial image set is performed. If the results of the consistency analysis show that the average visual attribute deviates from the fill attribute, then the preliminary cause of the difference pixel set is associated with process deviations related to color and tone reproduction. Based on the spatial location, preliminary cause, and corresponding design layer information of the differential pixel set, assign a corresponding defect type identifier to the differential pixel set; Based on the defect type identifier and the detailed cause information of the difference pixel set, a structured defect report of the quick-printed graphic product is generated.

8. The method for automatic identification and repair of image defect data in quick printing as described in claim 1, characterized in that, Based on the defect type and defect region category in the structured defect report, the initial image set is reconstructed to obtain the repaired image of the printed material, including: Identify the defect type, image / text region category, and spatial extent of the defect region in the initial image set from the defect region in the structured defect report; For defect types related to contour reconstruction, based on the path information of the corresponding design layer in the benchmark description set, the pixels within the defect range of the initial image set are reconstructed at the edges to obtain the contour reconstruction region of the initial image set. For defect types related to color and tone reproduction, based on the fill attribute of the corresponding design layer in the benchmark description set and the normal pixel color distribution around the defect area in the initial image set, the pixels in the defect area are visually consistent and reconstructed to obtain the color reconstruction area of ​​the initial image set. The restored image of the graphic print is generated by integrating the contour reconstruction region, the color reconstruction region, and the unmodified original image region in the initial image set.

9. The method for automatic identification and repair of image defect data in quick printing as described in claim 1, characterized in that, The step of verifying the consistency between the repaired image and the baseline description set to generate a repair report for the printed material includes: Spatially align the repaired image and the reference description set, and extract the repaired image feature description of the aligned repaired image and the reference feature description of the aligned reference description set; The average color, texture complexity, and edge sharpness of the regions corresponding to the baseline feature description in the restored image feature description are summarized into the region visual features of the restored image feature description; The visual attributes in the region visual features and the baseline feature description are compared item by item to obtain the visual deviation result of the repaired image feature description. The background regions in the restored image that are not covered by the benchmark description set are identified, and the naturalness of the transition in color and texture and the overall coordination of the background regions are analyzed to obtain the continuity verification results of the restored image. Based on the combined results of the visual deviation and the continuity verification, a repair report for the printed graphic product is generated.

10. An automatic identification and repair system for image defect data in rapid printing, characterized in that, The system for implementing the automatic identification and repair method for image defect data in quick printing as described in claim 1 includes: The data acquisition module is used to acquire optical scan image data and digital design source data of the graphic quick prints to obtain the initial image set and reference description set of the graphic quick prints. The expected printed image simulation module is used to classify the pixels in the initial image set into different graphic regions according to the object attributes in the reference description set, and to perform simulation rendering on the reference description set according to the graphic regions to obtain the expected printed image of the graphic quick print. The defect area identification module is used to perform multi-channel pixel-level comparison between the expected printed image and the initial image set to obtain the suspected defect areas of the quick-printed graphic product. The defect report generation module is used to combine the layer and path information in the benchmark description set to analyze the cause and defect type of the difference pixel set in the suspected defect area, and obtain a structured defect report of the graphic quick print. The image restoration module is used to perform targeted content reconstruction on the initial image set based on the defect type and defect area category in the structured defect report, so as to obtain the restored image of the graphic print. The repair report generation module is used to verify the consistency between the repaired image and the baseline description set in order to generate a repair report for the printed image.

Images