Image processing method applied to printed matter surface color difference detection

Abstract

The invention discloses an image processing method applied to printed matter surface color difference detection, which relates to the technical field of image processing, and comprises the following steps: acquiring a digital image of a printed matter to be detected, performing illumination non-uniformity correction, and converting the corrected digital image into a CIELAB color space; performing region segmentation on the digital image based on double constraint conditions of color gradient and texture boundary to generate a detection region graph; feature parameters are extracted based on the detection area graph, and a feature data set is generated; establishing a dynamic reference model by utilizing process parameters and material characteristics of the printed matter; and calculating the distance between the feature data set and the dynamic reference model, identifying color difference regions and generating color difference scores, and screening and grading the color difference regions according to the color difference scores. According to the method, a multi-layer image pyramid and homomorphic filtering combined illumination correction technology is adopted, and an adaptive weight fusion mechanism based on image features is introduced, so that the illumination nonuniformity is effectively eliminated while the definition of printing details is kept.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to an image processing method for detecting color differences on printed matter surfaces. Background Art

[0002] Color quality control of printed materials is a crucial component of the printing industry, directly impacting the visual quality and market value of products. Currently, color difference detection for printed materials relies primarily on manual visual inspection or traditional colorimeter measurement. Manual visual inspection is susceptible to factors such as subjective judgment, fatigue, and ambient lighting, resulting in poor consistency and repeatability of test results. Traditional colorimeter measurement, while offering high accuracy, is limited to point measurements and cannot capture the overall color distribution across the printed surface.

[0003] With the development of machine vision technology, image processing-based color difference detection methods for printed products have gradually emerged. However, existing image processing methods still have shortcomings in illumination correction, region segmentation, and feature extraction. In particular, they lack comprehensive consideration of printing process parameters and material properties, making it difficult to dynamically adjust detection standards based on the characteristics of different batches of printed products. Moreover, color difference evaluation methods are overly simplistic and fail to fully utilize multidimensional feature information, making it difficult to meet the differentiated detection needs of different regions. Summary of the Invention

[0004] The present invention provides an image processing method for detecting color differences on printed surfaces, which is used to solve technical problems in the prior art such as insufficient illumination correction, inaccurate area segmentation, fixed detection standards, and single color difference evaluation.

[0005] In view of this, a first aspect of the present invention provides an image processing method for detecting color differences on the surface of printed matter, comprising:

[0006] Acquire a digital image of the printed product to be inspected, perform illumination non-uniformity correction, and convert the corrected digital image into the CIELAB color space;

[0007] In CIELAB color space, the digital image is segmented based on the dual constraints of color gradient and texture boundary to generate a detection area map.

[0008] Extract feature parameters based on the detection area map and generate a feature data set;

[0009] Use printing process parameters and material characteristics to establish a dynamic reference model;

[0010] Calculate the distance between the feature dataset and the dynamic reference model, identify the color difference area and generate a color difference score, and screen and grade the color difference area according to the color difference score;

[0011] The filtered and graded color difference areas are optimized and the color difference detection results are output.

[0012] Optionally, performing illumination non-uniformity correction includes:

[0013] Perform Gaussian pyramid decomposition on the acquired digital image to construct a multi-layer image pyramid;

[0014] Extract the brightness component of each layer of the image in the multi-layer image pyramid and perform logarithmic transformation on the brightness component;

[0015] Convert the transformed luminance component to the frequency domain, apply a homomorphic filter to suppress low-frequency components, and perform an inverse transform;

[0016] Adjust the processed layers of images to the original digital image size, determine the weight coefficient of each layer of images based on image features, and perform multi-scale fusion;

[0017] The fused brightness information is combined with the color information of the original digital image to generate a digital image after illumination correction.

[0018] Optionally, performing region segmentation on the digital image based on the dual constraints of color gradient and texture boundary includes:

[0019] Extract multi-channel gradients of digital images in CIELAB color space and generate comprehensive color gradient maps;

[0020] Based on the digital image, a local gray-level co-occurrence matrix is ​​constructed, and a Gaussian mixture model is used for texture clustering. After morphological processing, a texture boundary map is generated.

[0021] Select the initial seed point based on the comprehensive color gradient map, use the texture boundary map as a constraint, execute the region growing algorithm, and generate the initial region segmentation result;

[0022] Analyze the frequency characteristics of printed dots in the initial region segmentation results, optimize the region boundaries based on the analysis results, and obtain a boundary optimized segmentation map;

[0023] Calculate the color statistical features and boundary features of each region in the boundary optimization segmentation map, merge adjacent regions that meet the preset merging conditions, and generate a merged segmentation map;

[0024] Assign a unique identifier to each region in the merged segmentation map and output a detection region map.

[0025] Optionally, the feature data set includes contour features, color features, texture features, and printing features.

[0026] Optionally, establishing a dynamic reference model includes:

[0027] Acquire and preprocess batch reference datasets and historical batch datasets; the batch reference datasets include standard sample feature datasets for each batch of printed products;

[0028] Perform cluster analysis on the standard sample feature data set to determine the feature distribution pattern and cluster center, and form the feature statistical structure;

[0029] Obtain the printing process parameters and material characteristic parameters corresponding to each batch, establish the mapping relationship between process parameters, material characteristics and feature statistics, and calculate the feature adjustment coefficient matrix;

[0030] Apply the feature adjustment coefficient matrix to adjust the feature statistical structure and generate the feature benchmark corresponding to each batch;

[0031] Based on the similarity analysis of the batch reference data set and the historical batch data set, combined with the characteristic benchmarks of each batch, the dynamic tolerance range is determined to form the color difference evaluation standard for each batch of printed products;

[0032] Integrate the characteristic benchmarks of each batch and the color difference evaluation standards of printed products to build a dynamic reference model.

[0033] Optionally, identifying the color difference region and generating the color difference score includes:

[0034] According to the batch identification of the printed product to be tested, the corresponding feature benchmark and printed product color difference evaluation standard are retrieved from the dynamic reference model;

[0035] Calculate the normalized Euclidean distance between each feature in the feature dataset and the corresponding feature in the feature benchmark to generate a feature distance matrix;

[0036] Based on the visual sensitivity and feature reliability index of each detection area, the weight coefficients of various feature distances are determined, and a weighted fusion method is used to generate a comprehensive distance metric to form a regional color difference quantization value;

[0037] Compare the regional color difference quantization value with the dynamic tolerance range in the print color difference evaluation standard, identify the color difference area that exceeds the tolerance range, and determine the color difference level according to the degree of excess;

[0038] Perform spatial clustering on the color difference areas and merge adjacent color difference areas of the same level;

[0039] Calculate the area ratio, boundary clarity, and severity of each color difference area, and determine the significance score of each color difference area based on the printed product quality specification requirements;

[0040] The significance score is used as the color difference score, and the color difference areas are sorted and graded to form the color difference area grading result.

[0041] Optionally, determining the weight coefficient of each layer of image based on the image feature includes:

[0042] Extract the brightness gradient amplitude features and local contrast features of each layer of image and perform normalization processing;

[0043] Calculate the comprehensive feature value based on the brightness gradient amplitude feature and the local contrast feature;

[0044] For different level images in a multi-layer image pyramid, set level weight factors;

[0045] The comprehensive eigenvalue is combined with the layer weight factor to generate the weight coefficient of each layer image.

[0046] A second aspect of the present invention provides an image processing system for detecting color differences on printed matter surfaces, comprising:

[0047] An illumination correction module is used to obtain a digital image of the printed product to be inspected, perform illumination non-uniformity correction, and convert the corrected digital image into the CIELAB color space;

[0048] A region segmentation module is used to perform region segmentation on a digital image based on the dual constraints of color gradient and texture boundary in the CIELAB color space to generate a detection region map;

[0049] A feature extraction module is used to extract feature parameters based on the detection area map and generate a feature data set;

[0050] Reference model building module, used to build a dynamic reference model using printing process parameters and material properties;

[0051] The color difference grading module is used to calculate the distance between the feature dataset and the dynamic reference model, identify the color difference area and generate a color difference score, and then screen and grade the color difference area according to the color difference score;

[0052] The optimization processing module is used to optimize the color difference areas after screening and classification and output the color difference detection results.

[0053] The beneficial effects of the present invention are as follows: the present invention adopts an illumination correction technology combining a multi-layer image pyramid and homomorphic filtering, and introduces an adaptive weight fusion mechanism based on image features, which effectively eliminates illumination unevenness while maintaining the clarity of printed details; proposes a dual-constraint segmentation method based on color gradient and texture boundary, combines printing dot frequency feature analysis and regional boundary optimization technology, improves the segmentation accuracy of complex patterns and gradient areas of printed products, and effectively reduces over-segmentation and under-segmentation phenomena; the dynamic reference model realizes precise adjustment of feature benchmarks through XGBoost regressor and Shapley value weight allocation, solves the problem that traditional fixed standard models are difficult to adapt to the differences in characteristics of different batches of printed products; constructs a color difference evaluation system that integrates visual perception characteristics, and realizes accurate quantitative evaluation of color difference through multi-dimensional feature distance calculation, adaptive weight fusion and dynamic threshold grading, thereby improving the accuracy of color difference detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0055] Figure 1 The figure is a flow chart of an image processing method for detecting color differences on the surface of printed matter.

[0056] Figure 2 The image correction flow chart of an image processing method applied to the detection of color differences on the surface of printed matter.

[0057] Figure 3 The figure shows an image region segmentation flow chart of an image processing method for detecting color differences on printed surfaces. DETAILED DESCRIPTION

[0058] In order to make the purpose, features, and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0059] Reference Figures 1 to 3 , is an embodiment of the present invention, which provides an image processing method for detecting color differences on printed surfaces. The flowchart of the method is shown in FIG. Figure 1 As shown, the method includes,

[0060] S1: Acquire a digital image of the printed product to be inspected and perform illumination non-uniformity correction, and convert the corrected digital image into the CIELAB color space.

[0061] In a specific embodiment of the present invention, the image correction flow chart is as follows: Figure 2 As shown, the following steps are included:

[0062] S1.1: Acquire a digital image of a printed product to be inspected, perform Gaussian pyramid decomposition on the acquired digital image, and construct a multi-layer image pyramid.

[0063] The first layer is the original digital image, and each subsequent layer image is obtained by performing Gaussian smoothing and downsampling on the previous layer. The number of pyramid layers N is determined adaptively according to the image resolution. If the image size is ,but .

[0064] In addition, during the acquisition process, it is necessary to ensure that the image resolution meets the detection accuracy requirements.

[0065] S1.2: Extract the brightness component of each layer of the image in the multi-layer image pyramid and perform logarithmic transformation on the brightness component.

[0066] It's important to note that when performing a logarithmic transformation, a small positive number (such as 0.001) must be added to the luminance component to prevent the logarithm from being invalid when the luminance value is zero. Extracting the luminance component and performing a logarithmic transformation helps compress the dynamic range and make the luminance distribution more uniform.

[0067] S1.3: Convert the transformed luminance component to the frequency domain, apply a homomorphic filter to suppress low-frequency components representing illumination non-uniformity, and perform an inverse transform.

[0068] In this embodiment, the frequency domain conversion uses a two-dimensional Fourier transform to efficiently separate the low-frequency illumination component (concentrated in the center of the frequency domain) from the high-frequency reflection component (distributed at the edge of the frequency domain); the inverse transform uses an inverse Fourier transform to ensure that the frequency domain filtering results are losslessly restored to the spatial domain.

[0069] In addition, the homomorphic filter The transfer function satisfies:

[0070]

[0071] in, is the frequency domain coordinate, and They are high-frequency gain and low-frequency gain respectively (high-frequency gain controls the enhancement strength of high-frequency components, and low-frequency gain controls the suppression strength of low-frequency components). is the slope constant, is the cutoff frequency (ranging from 0.05 to 0.2 times the image size). Homomorphic filtering was chosen because printed images conform to the illumination-reflection imaging model. Homomorphic filtering transforms this multiplicative model into an additive model through logarithmic transformation, facilitating frequency domain separation. It can independently control high-frequency reflection components (enhancing printed details) and low-frequency illumination components (suppressing uneven lighting), outperforming traditional filtering methods.

[0072] S1.4: The processed images of each layer are adjusted to the original digital image size by upsampling, and the weight coefficients of each layer of images are determined based on the image features to perform multi-scale fusion.

[0073] Image features include brightness gradient amplitude and local contrast. Specifically, brightness gradient amplitude is obtained by using the Sobel operator to calculate the horizontal and vertical gradients, respectively, and then taking the modulus. Local contrast is obtained by calculating the ratio of the standard deviation to the mean of the brightness components within a neighborhood. Ideally, the combination of brightness gradient amplitude and local contrast can comprehensively characterize image structural information: brightness gradient amplitude effectively captures edges and details, while local contrast accurately reflects changes in illumination uniformity. The weighted combination of the two ensures the clarity of printed pattern edges while suppressing interference caused by uneven illumination.

[0074] Furthermore, determining the weight coefficients of each layer of images based on image features includes: extracting the brightness gradient amplitude features and local contrast features of each layer of images, performing normalization processing, and mapping their value range to the interval [0, 1]; calculating the comprehensive feature value using a weighted method, and in this embodiment, assigning a larger weight to the gradient amplitude feature; setting a layer weight factor for different layers of images in a multi-layer image pyramid, generally assigning a larger basic weight to lower layers and appropriately reducing the basic weight to higher layers; and combining the comprehensive feature value with the layer weight factor to generate the weight coefficient of each layer of images.

[0075] Furthermore, in the multi-scale fusion stage, the images of each layer are subjected to pixel-level weighted fusion according to the calculated weight coefficients.

[0076] Through the above steps, the present invention effectively eliminates illumination non-uniformity in printed images. Compared with traditional global illumination correction methods, this method achieves a more uniform illumination distribution while maintaining edge clarity, laying a solid foundation for subsequent color difference detection.

[0077] S1.5: Combine the fused brightness information with the color information of the original digital image to generate a digital image after illumination correction.

[0078] It should be noted that this step is designed to solve the color distortion problem that may be caused during the lighting correction process, ensuring that the color authenticity is maintained while correcting the lighting.

[0079] S1.6: Convert the corrected digital image to the CIELAB color space.

[0080] Furthermore, the conversion to the CIELAB color space is accomplished by first converting the image from the RGB color space to the XYZ color space, and then converting from the XYZ color space to the CIELAB color space.

[0081] S2: In the CIELAB color space, region segmentation is performed on the digital image based on the dual constraints of color gradient and texture boundary to generate a detection region map.

[0082] In a specific embodiment of the present invention, the image region segmentation flow chart is as follows: Figure 3 As shown, the following steps are included:

[0083] S2.1: Extract multi-channel gradients of digital images in CIELAB color space and generate a comprehensive color gradient map.

[0084] It should be noted that the purpose of extracting multi-channel gradients of digital images in the CIELAB color space is to comprehensively capture color variation information. Specifically, the gradient amplitudes of the L, a, and b channels are calculated separately. The Sobel operator is used to calculate the gradients in the horizontal and vertical directions. The gradient characteristics of the three channels are comprehensively considered and assigned different weights (typically, the L channel gradient is given a weight of 0.4, and the a and b channel gradients are each given a weight of 0.3). The weighted gradient amplitudes of the three channels are then fused to generate a comprehensive color gradient map.

[0085] S2.2: Based on the digital image, a local gray-level co-occurrence matrix is ​​constructed, and a Gaussian mixture model is used for texture clustering. After morphological processing, a texture boundary map is generated.

[0086] Specifically, the luminance channel L is extracted from the digital image in the CIELAB color space, and the image is divided into multiple overlapping local regions; for each local region, the grayscale co-occurrence matrix in multiple directions is calculated, including four main directions: horizontal, vertical, diagonal and anti-diagonal, and four texture statistical features, namely contrast, correlation, energy and homogeneity, are extracted for each direction to form a local texture feature vector; the Gaussian mixture model is applied to the texture feature vector for unsupervised clustering to generate a texture category label map; morphological processing is performed on the texture category label map, including edge extraction, thinning and connection operations, to remove noise and isolated points, fill broken boundaries, and generate a texture boundary map with closed boundaries.

[0087] Preferably, by constructing a local gray-level co-occurrence matrix and Gaussian mixture model clustering, the differences in texture features of different regions of printed matter can be effectively identified, which is particularly suitable for printed matter regions with special texture structures such as dots, lines and fixed patterns.

[0088] S2.3: Select the initial seed point based on the comprehensive color gradient map, use the texture boundary map as a constraint, execute the region growing algorithm, and generate the initial region segmentation result.

[0089] Furthermore, threshold processing is performed on the comprehensive color gradient map generated in S2.1, and areas with gradient values ​​lower than the gradient selection threshold are identified as candidate seed areas, where the gradient selection threshold is determined according to the color complexity of the printed product; within each candidate seed area, the pixel point with the highest color uniformity and texture consistency index is selected as the initial seed point; the texture boundary map is converted into a constraint matrix, and the pixels at the texture boundary position are set as region growth blocking points; starting from each initial seed point, adaptive threshold region growth is performed, and a color similarity threshold is set, where it is dynamically adjusted according to the color variance of the local area to adapt to the color change characteristics of different printed areas; when the region growth process encounters a texture boundary constraint or the color difference between pixels exceeds the color similarity threshold, the expansion is stopped, and the growth process of a single region is completed; post-processing is performed on all generated regions, including small region merging and hole filling, to eliminate fragmented regions and generate initial region segmentation results.

[0090] Optimally, a region growing algorithm that combines color gradients with texture boundary constraints improves segmentation accuracy for gradient transitions and dot-changing regions in printed materials, while reducing over-segmentation. Furthermore, an adaptive threshold mechanism enables the algorithm to intelligently adapt to the local characteristics of different color regions, effectively addressing segmentation challenges posed by the unique dot structure and color transitions of printed materials.

[0091] S2.4: Analyze the printing dot frequency characteristics in the initial region segmentation result, optimize the region boundary based on the analysis result, and obtain a boundary optimized segmentation map.

[0092] Among them, the printing dot frequency characteristics include dot density, dot arrangement pattern and dot size distribution.

[0093] Specifically, the boundary area of ​​the initial region segmentation result is extracted; the dot feature differences of the areas on both sides of the boundary are detected, the dot density difference rate, arrangement pattern consistency and size distribution similarity are calculated, and the comprehensive feature difference index is determined based on these three indicators; for the boundary segment whose comprehensive feature difference index is lower than the preset feature difference judgment threshold, a boundary repositioning operation is performed to adjust the boundary to the nearest maximum position of the dot density gradient; for the boundary segment whose comprehensive feature difference index is higher than the preset feature difference judgment threshold, the boundary smoothness is improved through boundary optimization processing; the optimized boundary is fused with the initial region segmentation result, the region boundary information is updated, and a boundary optimized segmentation map is formed.

[0094] Among them, the feature difference judgment threshold is determined according to the type of printed matter and the detection accuracy requirements.

[0095] S2.5: Calculate the color statistical features and boundary features of each region in the boundary optimization segmentation map, merge adjacent regions that meet the preset merging conditions, and generate a merged segmentation map.

[0096] Among them, color statistical features include the average CIELAB value, standard deviation, color histogram distribution, etc. of pixels in the area, and boundary features include boundary gradient amplitude, directional consistency, and boundary curvature, etc.; color difference values ​​are calculated based on the color statistical features of adjacent areas, and boundary intensity values ​​are calculated based on the boundary features. These two values ​​are used to determine whether adjacent areas meet the preset merging conditions.

[0097] Furthermore, the preset merging condition states that adjacent regions are merged when the color difference between them is less than a color difference threshold and the boundary strength is less than a boundary strength threshold. Furthermore, the color difference threshold and boundary strength threshold can be adaptively adjusted based on factors such as the region's size, shape, and the characteristics of the surrounding area.

[0098] Optionally, for areas that may contain special materials such as metallic ink or fluorescent ink (pre-identified by detecting high brightness, high saturation or abnormal reflection characteristics), local reflection characteristic analysis is used to add a reflection uniformity index to the merging judgment conditions to prevent the special material area from being erroneously merged with the ordinary printing area.

[0099] It should be noted that if the preset merging conditions are not met, there is no need to merge and it is used as a merged segmentation map.

[0100] S2.6: Assign a unique identifier to each region in the merged segmentation map and output a detection region map.

[0101] S3: Extract feature parameters based on the detection area map and generate a feature data set.

[0102] In a specific embodiment of the present invention, step S3 includes the following steps:

[0103] S3.1: Based on the detection area map, extract the geometric parameters of each detection area and generate contour features.

[0104] Among them, the geometric parameters include area, perimeter and shape complexity. The shape complexity is expressed by the ratio of the square of the perimeter to the area.

[0105] S3.2: In the CIELAB color space, extract the color statistical distribution parameters of each detection area, calculate the color space distribution parameters, and generate color features.

[0106] Among them, the color statistical distribution parameters include the mean and standard deviation of the brightness L channel and the chromaticity a and b channels. The color space distribution parameters include color aggregation and color boundary sharpness. The color aggregation is quantified by the clustering degree of similar color pixels in the detection area, and the color boundary sharpness is quantified by calculating the color gradient amplitude at the boundary of the detection area.

[0107] S3.3: Apply texture feature extraction technology to extract structural texture parameters and statistical texture parameters and construct texture features.

[0108] Specifically, based on the local binary pattern feature map, the texture directionality and roughness of each detection area are analyzed to obtain the structural texture parameters; the gray level co-occurrence matrix is ​​used to calculate the texture contrast and homogeneity of each detection area to obtain the statistical texture parameters.

[0109] S3.4: Apply printing feature analysis technology to extract printing process parameters and printing quality parameters and construct printing features.

[0110] Furthermore, the dot distribution density, dot arrangement consistency and dot size uniformity are extracted to obtain printing process parameters, where the dot distribution density is calculated by the precise number of dots per unit area, the dot arrangement consistency is quantified by the variability of the dot spacing, and the dot size uniformity is characterized by the standard deviation of the dot size within the area; the color contrast between each detection area and the adjacent area is calculated to obtain printing quality parameters.

[0111] S3.5: Screen and normalize the extracted contour features, color features, texture features, and printing features to generate a structured feature dataset.

[0112] It should be noted that screening includes applying feature selection algorithms to evaluate the importance of various extracted features, retaining features that contribute significantly to color difference detection, and eliminating redundant features; normalization processing includes mapping the retained features to a uniform numerical range and organizing them by region identifiers.

[0113] The various features in the feature dataset are optimally combined based on the detection objective. Contour features primarily describe the spatial morphology of the color difference region, color features reflect the color distribution within the region, texture features reflect the surface structure of the region, and print features specifically characterize the quality of the printing process. This multi-dimensional, coordinated representation of features comprehensively reflects the color differences on the printed surface, providing a reliable data foundation for subsequent color difference evaluation.

[0114] S4: Use the printing process parameters and material properties to establish a dynamic reference model.

[0115] In a specific embodiment of the present invention, step S4 includes the following steps:

[0116] S4.1: Acquire and preprocess the batch reference dataset and historical batch dataset.

[0117] The batch reference dataset includes the characteristic datasets of standard samples for each batch of printed products. These datasets are derived from qualified standard printed products within the batch using the same feature extraction method as the printed products to be tested. The historical batch datasets are the characteristic datasets of standard samples from all previous batches stored in the system database. The batch reference dataset is collected during production quality control, while the historical batch dataset is retrieved through database system retrieval.

[0118] Furthermore, for new batches of printed products being tested for the first time, when standard samples are unavailable, the system selects batch data with the most similar process and material characteristics from historical batch datasets based on the batch's printing process parameters and material characteristics to construct a standard sample feature dataset for that batch. Similarity is determined by calculating the weighted distance between printing process information (such as printing equipment type, ink formulation, and printing speed) and material characteristics (such as paper type, surface treatment, and base whiteness). This similarity matching mechanism, based on process parameters and material characteristics, overcomes the problem of a lack of standard samples during the initial testing of new batches of printed products and ensures the continuous availability of the inspection system.

[0119] S4.2: Perform cluster analysis on the standard sample feature data set to determine the feature distribution pattern and cluster centers and form the feature statistical structure.

[0120] Specifically, the contour, color, texture, and printing features in the standard sample feature dataset were normalized to eliminate dimensional differences. Correlation analysis and dimensionality reduction were performed on the normalized features to form a simplified feature vector. Based on the simplified feature vector, a Gaussian mixture model was used for clustering, and the optimal number of clusters was automatically determined using the Bayesian Information Criterion. The Gaussian mixture model clustering method was found to be more effective in handling the non-normal distribution of printing features. Initial clustering results were used to parameterize the clustering results, analyze the degree of cluster dispersion and the distance relationship between clusters, and evaluate the clustering quality. A topological structure of each feature was constructed with cluster centers as nodes, establishing a hierarchical organization of the feature space. The cluster topology was batch-labeled using batch information, a batch feature matrix was established, and a bidirectional mapping relationship between batch and feature distribution was constructed. The cluster analysis results and batch mapping relationships of all feature categories were integrated to generate a complete feature statistical structure. Clustering using a Gaussian mixture model and the establishment of a bidirectional mapping relationship between batch and feature distribution effectively addressed the modeling challenges of the non-normal distribution of printing features and improved the accuracy of the representation of the feature statistical structure.

[0121] S4.3: Obtain the printing process parameters and material characteristic parameters corresponding to each batch, establish a mapping relationship between the process parameters, material characteristics and feature statistics, and calculate the feature adjustment coefficient matrix.

[0122] Furthermore, the printing process parameters and material characteristic parameters of each batch are obtained, the obtained parameters are normalized, and a parameter matrix is ​​constructed. Based on the characteristic statistical structure formed in S4.2, the characteristic center value, distribution range, and dispersion corresponding to each batch are extracted to construct a characteristic statistic matrix. The XGBoost regressor is used to establish a mapping function model from the parameter matrix to the characteristic statistic matrix, and Shapley value weight distribution is introduced. The parameter sensitivity of the established mapping function model is calculated to form a sensitivity matrix to quantify the influence of each parameter on different characteristic statistics. Combined with the actual variation range of the parameters and the allowable fluctuation range of production, the sensitivity matrix is ​​converted into a characteristic adjustment coefficient matrix. By introducing the XGBoost regressor with Shapley value weight distribution to establish a mapping relationship, the influence of process parameters and material characteristics on characteristic statistics is accurately quantified, providing a reliable basis for the dynamic adjustment of characteristic benchmarks.

[0123] S4.4: Apply the feature adjustment coefficient matrix to adjust the feature statistical structure and generate feature benchmarks corresponding to each batch.

[0124] Furthermore, according to the printing process parameters and material characteristic parameters of the current batch, the corresponding adjustment coefficients are extracted from the feature adjustment coefficient matrix; the feature adjustment coefficients are applied to each cluster center and its distribution range in the feature statistical structure; adjustment calculations are performed on the statistical parameters of color features, texture features, contour features and printing features respectively; based on the adjusted feature cluster centers and distribution ranges, the batch feature statistical structure is reconstructed; the reconstructed feature statistical structure is verified and saved as the feature benchmark of the current batch.

[0125] S4.5: Based on the similarity analysis of the batch reference data set and the historical batch data set, combined with the characteristic benchmarks of each batch, determine the dynamic tolerance range and form the color difference evaluation standard for each batch of printed products.

[0126] Furthermore, the similarity index between the feature benchmark of the current batch and the feature benchmark of each batch in the historical batch data set is calculated, and the historical batch data closest to the current batch is screened out based on the similarity index; the feature difference distribution of qualified products and unqualified products in the screened historical batches is analyzed, and the initial tolerance range of each feature is calculated based on the feature benchmark of the current batch; the initial tolerance range is corrected according to the stability of the printing process and the material characteristics to generate a dynamic tolerance range.

[0127] S4.6: Use a hash table structure with batch identifier as the key to establish a batch indexing mechanism, integrate the characteristic benchmarks of each batch and the color difference evaluation standards of printed products, and build a dynamic reference model.

[0128] It's important to note that the dynamic reference model can quickly extract the corresponding batch's characteristic benchmarks and print color difference evaluation standards based on the batch identification of the printed product being tested, for comparison and evaluation with the characteristic dataset of the printed product being tested. By establishing a precise mapping relationship between process parameters, material properties, and characteristic statistics, this technical solution enables dynamic adjustment of the testing standard. This overcomes the difficulty of traditional fixed standard models in adapting to the differences in characteristics of different batches of printed products, significantly improving the adaptability and accuracy of color difference detection.

[0129] S5: Calculate the distance between the feature dataset and the dynamic reference model, identify the color difference area and generate a color difference score, and screen and grade the color difference area according to the color difference score.

[0130] In a specific embodiment of the present invention, step S5 includes the following steps:

[0131] S5.1: Retrieve the corresponding feature benchmark and printed product color difference evaluation standard from the dynamic reference model based on the batch identification of the printed product to be tested.

[0132] S5.2: Calculate the normalized Euclidean distance between each feature in the feature dataset and the corresponding feature in the feature benchmark to generate a feature distance matrix.

[0133] It should be noted that when calculating the normalized Euclidean distance, the Z-score standardization method is first used to unify and normalize the dimensions of various features to ensure comparability of features of different ranges and magnitudes. For each of the four major feature categories—contour, color, texture, and print—the Euclidean distance to the feature baseline is calculated to construct a feature distance vector. Furthermore, considering the potential correlation between features, a Mahalanobis distance correction mechanism is introduced to reduce the impact of feature redundancy on distance calculation. For each detection area, the four feature distance vectors are organized into a feature distance matrix. This Mahalanobis distance correction mechanism effectively addresses the problem of inter-feature correlation interfering with distance calculation, improving the accuracy of feature distance measurement.

[0134] S5.3: Based on the visual sensitivity and feature reliability indicators of each detection area, determine the weight coefficients of various feature distances, use the weighted fusion method to generate a comprehensive distance metric, and form a regional color difference quantization value.

[0135] Specifically, the system obtains the visual sensitivity characteristics of each inspection area (including area, location, boundary complexity, and color contrast), and calculates the visual importance index of each inspection area based on human visual characteristics. It then performs a reliability assessment on four feature categories: contour, color, texture, and printing features, quantifying their scores based on three dimensions: feature stability, feature discrimination, and feature consistency, to generate a feature reliability score. Based on the type and process parameters of the printed product being inspected, a matching basic weight configuration is retrieved from a pre-set weight template library and used as the initial weight allocation scheme. The basic weight configuration is adaptively adjusted based on the visual importance index and feature reliability score of each inspection area to generate a final feature weight coefficient matrix, ensuring that the sum of the weight coefficients for each inspection area is 1. A weighted fusion is performed on the feature distance matrix and the weight coefficient matrix to generate a regional comprehensive distance map. The comprehensive distance map is then nonlinearly mapped based on printing process standards and batch characteristics to align its distribution with the human eye's perception of color difference, resulting in the final regional color difference quantification value. This adaptive weight allocation mechanism, based on visual sensitivity and feature reliability, enables accurate assessment of the feature importance of different inspection areas and significantly improves the consistency of color difference detection results with human perception.

[0136] S5.4: Compare the regional color difference quantification values ​​with the dynamic tolerance range in the printed product color difference evaluation standard, identify the color difference areas that exceed the tolerance range, and determine the color difference level based on the degree of excess.

[0137] Determining the color difference grade based on the degree of excess includes: calculating the deviation ratio between the color difference quantization value of each detection area and the upper limit of the dynamic tolerance range; establishing a five-level color difference grading standard based on the deviation ratio, including slight color difference, general color difference, obvious color difference, severe color difference, and extremely severe color difference, and using an adaptive threshold method to determine the critical value of each grade; dynamically adjusting the critical value of each grade in combination with the quality control requirements of the printing process, and taking into account the visual significance of the color difference area for secondary correction, thus forming a color difference grading mechanism. By establishing a color difference grading standard with an adaptive threshold and performing secondary correction based on visual significance, the problem that the traditional fixed threshold method is difficult to adapt to the quality requirements of different printed products is solved.

[0138] S5.5: Perform spatial clustering on the color difference regions and merge adjacent color difference regions of the same level.

[0139] It should be noted that spatial clustering uses an adaptive clustering algorithm based on region growing. Specifically, a connectivity graph is first established based on the spatial adjacency of color difference regions. The similarity of color difference features is then calculated for adjacent color difference regions of the same level. Regions are merged when the similarity exceeds a preset threshold, where the merging threshold is dynamically adjusted based on the color difference level. This method effectively reduces the fragmentation of color difference regions, improving the practicality of detection results while maintaining the accuracy of color difference region boundaries.

[0140] S5.6: Calculate the area percentage, boundary clarity, and severity of each color difference area, and determine the significance score for each color difference area based on the requirements of the printed product quality specification.

[0141] Specifically, for each color difference region, the ratio of its area to the area of ​​the entire printed product is calculated, an area ratio scoring function is established, and an area ratio score is generated; the boundary contour of the color difference region is extracted, the gradient amplitude and directional consistency of the edge pixels are analyzed, the boundary clarity is quantified, and a boundary clarity score is generated; the color difference level of the color difference region and the uniformity of the color difference distribution within the region are combined to evaluate the perceived significance of the color difference and generate a color difference severity score; according to the application scenario of the printed product, the corresponding color difference evaluation standard is retrieved from the quality specification database, and the weight coefficients of different scoring indicators (area ratio, boundary clarity, color difference severity) are determined; the area ratio score, boundary clarity score and color difference severity score are weighted and combined to generate a comprehensive significance score for each color difference region.

[0142] S5.7: Use the significance score as the color difference score, sort and grade the color difference areas, and generate a color difference area grading result.

[0143] Furthermore, all color difference areas are sorted from high to low according to their significance scores to determine the priority processing order of each color difference area. Then, a dynamic threshold method is used to divide the color difference areas into three levels: high significance, medium significance, and low significance. The threshold is adaptively determined based on the statistical distribution of the significance scores of the current batch of printed products. Combining the color difference grades (slight color difference, general color difference, obvious color difference, severe color difference, and extremely severe color difference) and significance levels obtained in S5.4, a two-dimensional grading matrix is ​​constructed to generate the final color difference area grading results.

[0144] Advantageously, the color difference identification and scoring method proposed in this invention establishes a comprehensive color difference quantitative evaluation system through innovative technologies such as multidimensional feature distance calculation, adaptive weight fusion, and dynamic threshold grading. This method not only considers objective color difference measurement indicators but also incorporates the human visual perception characteristics and printing quality standards. This ensures that color difference detection results are more in line with practical application needs while also being highly adaptable and scalable.

[0145] S6: Optimize the color difference areas after screening and grading, and output the color difference detection results and evaluation report.

[0146] Among them, the optimization processing includes: applying morphological filtering algorithms to optimize the boundaries of color difference areas to improve edge smoothness and continuity; removing color difference areas with an area smaller than a preset threshold and a low significance score to reduce the impact of detection noise; classifying and labeling color difference areas based on printing process characteristics and color difference distribution patterns to distinguish between systematic color difference and random color difference; generating color difference detection results and evaluation reports, and providing color difference cause analysis and improvement suggestions.

[0147] Furthermore, this embodiment also provides an image processing system for detecting color differences on printed product surfaces, comprising: an illumination correction module for acquiring a digital image of a printed product to be detected, performing illumination non-uniformity correction, and converting the corrected digital image into a CIELAB color space; a region segmentation module for performing region segmentation on the digital image based on the dual constraints of color gradient and texture boundary in the CIELAB color space to generate a detection region map; and a feature extraction module for extracting feature parameters based on the detection region map to generate a feature data set.

[0148] The reference model construction module is used to establish a dynamic reference model using the printing process parameters and material characteristics; the color difference grading module is used to calculate the distance between the feature data set and the dynamic reference model, identify the color difference area and generate a color difference score, and screen and grade the color difference area according to the color difference score; the optimization processing module is used to optimize the color difference area after screening and grading, and output the color difference detection results and evaluation report.

[0149] In summary, the present invention adopts an illumination correction technology that combines a multi-layer image pyramid with homomorphic filtering, and introduces an adaptive weight fusion mechanism based on image features, which effectively eliminates illumination unevenness while maintaining the clarity of printed details; proposes a dual-constraint segmentation method based on color gradient and texture boundary, combined with printing dot frequency feature analysis and regional boundary optimization technology, improves the segmentation accuracy of complex patterns and gradient areas of printed products, and effectively reduces over-segmentation and under-segmentation phenomena; the dynamic reference model realizes precise adjustment of feature benchmarks through XGBoost regressor and Shapley value weight allocation, which solves the problem that traditional fixed standard models are difficult to adapt to the differences in characteristics of different batches of printed products; constructs a color difference evaluation system that integrates visual perception characteristics, and realizes accurate quantitative evaluation of color difference through multi-dimensional feature distance calculation, adaptive weight fusion and dynamic threshold grading, thereby improving the accuracy of color difference detection.

[0150] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. An image processing method for detecting color differences on printed surfaces, characterized in that: include: Acquire a digital image of the printed product to be inspected, perform illumination non-uniformity correction, and convert the corrected digital image into the CIELAB color space; In the CIELAB color space, performing region segmentation on the digital image based on dual constraints of color gradient and texture boundary to generate a detection region map; Extracting feature parameters based on the detection area map to generate a feature data set; Use printing process parameters and material characteristics to establish a dynamic reference model; Calculating the distance between the feature dataset and the dynamic reference model, identifying color difference areas and generating color difference scores, and screening and grading the color difference areas according to the color difference scores; The filtered and graded color difference areas are optimized and the color difference detection results are output.

2. The image processing method for detecting color differences on printed matter surfaces according to claim 1, characterized in that: The correction of illumination non-uniformity comprises: Perform Gaussian pyramid decomposition on the acquired digital image to construct a multi-layer image pyramid; extracting a brightness component of each layer of the image in the multi-layer image pyramid, and performing a logarithmic transformation on the brightness component; Convert the transformed luminance component to the frequency domain, apply a homomorphic filter to suppress low-frequency components, and perform an inverse transform; Adjust the processed layers of images to the original digital image size, determine the weight coefficient of each layer of images based on image features, and perform multi-scale fusion; The fused brightness information is combined with the color information of the original digital image to generate a digital image after illumination correction.

3. The image processing method for detecting color differences on printed matter surfaces according to claim 1, characterized in that: The performing region segmentation on the digital image based on the dual constraints of color gradient and texture boundary comprises: Extract multi-channel gradients of digital images in CIELAB color space and generate comprehensive color gradient maps; constructing a local gray-level co-occurrence matrix based on the digital image, performing texture clustering using a Gaussian mixture model, and generating a texture boundary map after morphological processing; Select the initial seed point based on the comprehensive color gradient map, use the texture boundary map as a constraint, execute the region growing algorithm, and generate the initial region segmentation result; Analyze the frequency characteristics of printed dots in the initial region segmentation results, optimize the region boundaries based on the analysis results, and obtain a boundary optimized segmentation map; Calculate the color statistical features and boundary features of each region in the boundary optimization segmentation map, merge adjacent regions that meet the preset merging conditions, and generate a merged segmentation map; Assign a unique identifier to each region in the merged segmentation map and output a detection region map.

4. The image processing method for detecting color differences on printed matter surfaces according to claim 1, characterized in that: The feature data set includes contour features, color features, texture features and printing features.

5. The image processing method for detecting color differences on printed matter surfaces according to claim 1, characterized in that: The establishment of the dynamic reference model comprises: Acquire and preprocess a batch reference dataset and a historical batch dataset; the batch reference dataset includes a standard sample feature dataset of each batch of printed products; Performing cluster analysis on the standard sample feature data set to determine feature distribution patterns and cluster centers to form a feature statistical structure; Obtain the printing process parameters and material characteristic parameters corresponding to each batch, establish the mapping relationship between process parameters, material characteristics and feature statistics, and calculate the feature adjustment coefficient matrix; Applying the feature adjustment coefficient matrix to adjust the feature statistical structure to generate feature benchmarks corresponding to each batch; Based on the similarity analysis of the batch reference data set and the historical batch data set, combined with the characteristic benchmarks of each batch, the dynamic tolerance range is determined to form the color difference evaluation standard for each batch of printed products; Integrate the characteristic benchmarks of each batch and the color difference evaluation standards of printed products to build a dynamic reference model.

6. The image processing method for detecting color differences on printed matter surfaces according to claim 1, characterized in that: The identifying of color difference areas and generating color difference scores comprises: According to the batch identification of the printed product to be tested, the corresponding feature benchmark and printed product color difference evaluation standard are retrieved from the dynamic reference model; Calculating the normalized Euclidean distance between each feature in the feature dataset and the corresponding feature in the feature benchmark to generate a feature distance matrix; Based on the visual sensitivity and feature reliability index of each detection area, the weight coefficients of various feature distances are determined, and a weighted fusion method is used to generate a comprehensive distance metric to form a regional color difference quantization value; Comparing the regional color difference quantization value with the dynamic tolerance range in the printed product color difference evaluation standard, identifying the color difference area exceeding the tolerance range, and determining the color difference level according to the degree of exceeding the tolerance range; Performing spatial clustering processing on the color difference regions, and merging adjacent color difference regions of the same level; Calculate the area ratio, boundary clarity, and severity of each color difference area, and determine the significance score of each color difference area based on the printed product quality specification requirements; The significance score is used as a color difference score, and the color difference areas are sorted and graded to form a color difference area grading result.

7. The image processing method for detecting color differences on printed matter surfaces according to claim 2, characterized in that: Determining the weight coefficients of each layer of images based on image features includes: Extract the brightness gradient amplitude features and local contrast features of each layer of image and perform normalization processing; Calculating a comprehensive feature value based on the brightness gradient amplitude feature and the local contrast feature; For different level images in a multi-layer image pyramid, set level weight factors; The comprehensive eigenvalue is combined with the layer weight factor to generate the weight coefficient of each layer image.

8. An image processing system for detecting color differences on printed surfaces, characterized in that: include: An illumination correction module is used to obtain a digital image of the printed product to be inspected, perform illumination non-uniformity correction, and convert the corrected digital image into the CIELAB color space; A region segmentation module is used to perform region segmentation on the digital image based on the dual constraints of color gradient and texture boundary in the CIELAB color space to generate a detection region map; A feature extraction module, configured to extract feature parameters based on the detection area map and generate a feature data set; Reference model building module, used to build dynamic reference models using printing process parameters and material properties; a color difference grading module, configured to calculate the distance between the feature dataset and the dynamic reference model, identify color difference regions and generate color difference scores, and screen and grade the color difference regions according to the color difference scores; The optimization processing module is used to optimize the color difference areas after screening and classification and output the color difference detection results.

Images