Printing quality monitoring method and system based on artificial intelligence

Abstract

The invention discloses a printing quality monitoring method and system based on artificial intelligence, and relates to the technical field of quality monitoring, and the method comprises the following steps: collecting a printing layout, and obtaining an initial feature; performing defect area identification according to the initial features to obtain defect information; performing color deviation analysis according to the defect information to obtain color deviation information; performing registration deviation calculation according to the color deviation information to obtain registration information; performing comprehensive quality evaluation according to the registration information to obtain quality evaluation information; and performing quality analysis according to the quality evaluation information to obtain a quality monitoring result. According to the method, accurate defect positioning is realized, the defect identification precision is improved, the missing detection and false detection rates are reduced, and the adaptability and universality of defect identification are enhanced.

Description

Technical Field

[0001] This invention relates to the field of quality monitoring technology, specifically to a printing quality monitoring method and system based on artificial intelligence. Background Technology

[0002] With the rapid development of the printing industry, especially the widespread application of high-precision color printing and complex graphic printing, the quality requirements for printed materials are becoming increasingly stringent. Traditional printing quality inspection mainly relies on manual visual inspection or simple machine vision inspection methods. Manual inspection is inefficient and the results are greatly affected by subjective factors. In recent years, artificial intelligence technology has made significant progress in image recognition, pattern analysis, and data processing, providing new possibilities for printing quality monitoring.

[0003] Existing technologies for defect detection have shortcomings: Existing technologies for defect detection typically employ traditional image segmentation or fixed template matching methods. These methods mainly rely on fixed thresholds to judge pixel anomalies or to compare the differences between the printed surface and the reference template to determine the defect area. This can easily lead to false detections or missed detections, and it is difficult to handle complex printed patterns or multiple defect types. In particular, in continuous printing batches, there are significant differences in defect types and manifestations, which traditional methods cannot adequately address. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a printing quality monitoring method and system based on artificial intelligence, thereby solving the problems mentioned in the background section.

[0005] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a printing quality monitoring method and system based on artificial intelligence, comprising the following steps: S1. Collect the printing plate layout to obtain initial features; S2. Based on the initial features, identify the defect area and obtain defect information; S3. Perform color deviation analysis based on defect information to obtain color deviation information; S4. Calculate the registration deviation based on the color deviation information to obtain the registration information; S5. Conduct a comprehensive quality evaluation based on the registration information to obtain quality evaluation information; S6. Conduct quality analysis based on quality evaluation information to obtain quality monitoring results.

[0006] To further optimize this technical solution, the printing plate acquisition in step S1 includes: By continuously acquiring images of the printed page using acquisition tools, complete and accurate information about the printed page is obtained, and initial features are extracted from this information.

[0007] To further optimize this technical solution, the defect area identification in step S2 includes: Based on the initial features obtained, defect classification and identification and defect severity quantification are achieved through region division and abnormal feature signal extraction, thereby obtaining defect information, achieving accurate defect location, and providing statistical indicators for the final quality score.

[0008] To further optimize this technical solution, the quantification of the degree of defect includes:

[0009] in: Defect severity; : Percentage of defective area; Normalized texture perturbation intensity; : The degree of color deviation during normalization; The degree of disorder in the normalized network nodes; : Weighting coefficient for the percentage of defective area; : Weighting coefficient for texture perturbation intensity; Weighting coefficient for the degree of color deviation; Weighting coefficients for the degree of branch network disorder; The severity of a defect is obtained by weighted summation of four types of defect influencing factors.

[0010] To further optimize this technical solution, the color deviation analysis in step S3 includes: Based on the obtained defect information, the pixels in the defect area are extracted. Through local color difference calculation and aggregation, the color difference performance of the defect area is evaluated, the ink color fluctuation within the area is revealed, and the color deviation information of the defect area is obtained.

[0011] To further optimize this technical solution, the local color difference calculation and aggregation includes: Pixels in the normal printing area are selected as a reference, and the color difference is compared pixel by pixel with those in the defective area. The defective area is divided into several micro-units, and the local color difference value of each micro-unit is calculated using pixel-level color difference data. The color difference data is then aggregated to obtain the regional color deviation.

[0012] To further optimize this technical solution, the calculation of registration deviation in step S4 includes: Using the obtained color deviation information, relatively uniform and clearly defined overlapping color areas are selected as the base area for registration analysis. The spatial relationship between each color plate is clarified, rotation and scaling deviations are identified, and registration information is obtained.

[0013] To further optimize this technical solution, the comprehensive quality evaluation in step S5 includes: Based on the registration information, combined with defect information and color deviation information, a comprehensive quality evaluation is conducted on each printed image, and the quality level is divided according to the scoring results to obtain quality evaluation information.

[0014] To further optimize this technical solution, the quality analysis in step S6 includes: Based on the obtained quality evaluation information, the data is classified and statistically analyzed to form trend analysis data. By comparing it with historical data, recurring high-risk areas, common defect types, and abnormal production patterns are identified, thereby achieving anomaly detection and pattern discovery and obtaining quality monitoring results.

[0015] This technical solution has been further optimized, including the following functional modules: The module includes a feature acquisition module, a defect identification module, a color analysis module, a registration analysis module, a quality scoring module, and a quality monitoring module.

[0016] In a second aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein the computer program instructions, when executed by the processor, implement the steps of the printing quality monitoring method and system based on artificial intelligence as described in the first aspect of the present invention.

[0017] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, they implement the steps of the printing quality monitoring method and system based on artificial intelligence as described in the first aspect of the present invention.

[0018] Compared with existing technologies, the present invention provides a printing quality monitoring method and system based on artificial intelligence, which has the following beneficial effects: This AI-based printing quality monitoring method and system achieves precise defect location through region division and abnormal feature signal extraction, improves defect identification accuracy, reduces missed and false detection rates, and enhances the adaptability and versatility of defect identification. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating an artificial intelligence-based printing quality monitoring method proposed in this invention. Figure 2 This is a schematic diagram of a printing quality monitoring system based on artificial intelligence proposed in this invention. Detailed Implementation

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0022] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0023] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0024] Example 1: Reference Figure 1 This is the first embodiment of the present invention, which provides a printing quality monitoring method based on artificial intelligence, including the following steps: S1. Collect the printed layout to obtain initial features.

[0025] In this embodiment, the acquisition of the printing plate includes: In the printing quality monitoring process, data acquisition is the foundation of the entire system. The surface information of the printed plate determines the accuracy of subsequent defect identification, color deviation calculation, and registration deviation analysis. If the initial data is incomplete or the resolution is insufficient, minute defects, subtle color differences, and slight registration misalignments will not be detected, leading to quality assessment biases. Since the printed plate typically contains multiple color overprinting areas, micro-patterns, and areas with different dot ratios, relying solely on single or low-resolution acquisition cannot guarantee the full capture of printing details. High-precision, high-resolution image acquisition is a prerequisite for ensuring the reliability of subsequent processing.

[0026] By continuously acquiring images of the printing plate using acquisition tools, complete and accurate information about the printing plate is obtained. Initial features are extracted from this information to ensure the accuracy and completeness of the data throughout the printing quality monitoring process, providing data support for subsequent steps.

[0027] The steps for acquiring a printing plate include: Selection of acquisition equipment: Use a high-speed industrial camera as the acquisition tool to ensure that the imaging resolution meets the requirements for detecting minute defects and fine textures, and can distinguish individual printing dots or tiny stains. The machine must be matched with the speed of the printing equipment to avoid image blurring or missed shots and ensure imaging continuity. Installation and positioning: Fix the industrial camera in a suitable position on the printing press so that the camera's field of view covers the entire printing plate, avoids image distortion or missing edge areas, and ensures uniform lighting to avoid shadows and reflections affecting image quality; Adjust imaging parameters: Set the camera frame rate, exposure time and light source brightness according to the printing speed and paper characteristics to ensure that every section of the page is captured without omission, the color is captured accurately, without color deviation or uneven brightness in some areas, and the captured image is clear and the brightness is uniform. Image acquisition: The printing plate is continuously scanned during the printing process to record the complete image of each printing area, including different color overprinting areas, edge areas and micro-pattern areas, and generate a high-resolution image sequence for each printing area to record the real-time printing status; Region segmentation and annotation: The acquired images are segmented, and key regions such as main patterns, dot areas and registration reference points are marked to provide target location for subsequent defect identification and registration deviation analysis, which is convenient for subsequent defect identification and deviation analysis. Feature extraction: Initial features are obtained from the acquired images, including color information (such as CIE-Lab channel values, used to calculate color difference and determine color deviation), dot ratio (used to evaluate printing uniformity and micro-defects), edge sharpness (measures the integrity of the pattern outline), texture entropy value (reflects pattern complexity and abnormal texture), and registration reference point position (provides accurate coordinates for subsequent registration deviation analysis), thus providing basic data for subsequent analysis; Data storage and management: Collected images and extracted initial features are stored according to print page number, batch, and collection time to ensure data integrity and traceability.

[0028] S2. Based on the initial features, identify the defect area to obtain defect information.

[0029] In this embodiment, the defect region identification includes: In the printing quality monitoring system, defect identification is one of the core links, because printing defects directly affect the product's appearance, readability and commercial value. In the actual printing process, problems such as missing prints, scratches, ghosting, dirt, ink accumulation and color spot peeling often occur. Different defect forms are obviously different, and most defects are small in area and scattered in location. Human eye detection has large errors and low efficiency. Traditional identification methods rely on simple thresholds or color comparison to judge, which easily misses minor flaws or mistakenly regards normal textures as defects.

[0030] Based on the initial features obtained, defect classification and identification and defect severity quantification are achieved through region division and abnormal feature signal extraction, thereby obtaining defect information, enabling accurate defect location, improving identification accuracy, significantly reducing missed detection and false detection rates, providing statistical indicators for the final quality score, realizing the differentiation of different types of defects and the selection of key analysis areas, enhancing the adaptability and versatility of defect identification, and improving the efficiency of subsequent processing steps.

[0031] Methods for defect region identification include: Establish feature input structure: Unify and organize the initial features collected by S1, including color channel values, dot ratio, texture complexity, edge sharpness and registration reference point position information, and use them as the basis for recognition input so that the subsequent analysis model can make a comprehensive judgment on each region based on multiple factors; Region segmentation processing: Printed surfaces typically contain various structures such as large color blocks, gradient areas, small text or patterns, and registration marks. Each region has a different defect manifestation. Based on the region labels generated by S1, the printed surface is segmented, making the identification process more targeted and improving the accuracy of judgment. Different regions are analyzed using different feature combinations. For example, text regions focus more on edge continuity, while dot regions focus more on texture uniformity. Abnormal feature signal extraction: Analyze color fluctuations, dot density changes, texture distribution anomalies, and contour continuity defects in each divided region. When a region exhibits abrupt brightness changes, small-area discontinuous structures, interrupted textures, or abnormal particle aggregation, it is marked as a candidate defect region for further in-depth identification. Defect classification and identification: Candidate defect areas are classified and judged by an artificial intelligence recognition network. Candidate defect areas are divided into categories such as missing print, scratches, color stains, ghosting, and ink accumulation, and defect type labels are generated. By focusing on the analysis of the differences in features such as color offset texture change patterns and structural damage morphology, it is possible to distinguish defect areas with similar morphology but different causes. Quantifying the severity of defects: Calculate the proportion of the defect area in the corresponding detection area, and combine the intensity of texture disturbance, the magnitude of color deviation and the degree of dot disorder to quantify the severity of defects. This transforms qualitative defect information into calculable and comparable values, enabling defect information to directly participate in the comprehensive scoring. Output structured defect information: Organize structured data and output defect information including defect category, location coordinates, area percentage, and severity.

[0032] Furthermore, the quantification of the degree of defect includes:

[0033] in: Defect severity reflects the severity of a single defect area and quantifies the impact of the defect on print quality. It ranges from 0 to 1, with a higher value indicating greater severity. Defect area percentage: This measures the proportion of the defective area within the inspection area, reflecting the size of the defect coverage. It ranges from 0 to 1. The larger the area percentage, the more significant the impact of the defect on the overall printing quality. Normalized texture perturbation intensity measures the deviation between the texture of the defective area and the texture of the normal area. The texture perturbation intensity is normalized to the range of 0 to 1 using the min-max normalization method. The larger the value, the greater the texture change and the more severe the defect. Normalized color deviation: This quantifies the severity of the overall color deviation in the defective area. Based on the maximum allowable color difference set by printing quality requirements, it normalizes the color deviation to a range of 0 to 1. The larger the value, the greater the deviation and the greater the impact on printing quality. Normalized dot disorder degree measures the deviation of the dot distribution in the defective area from the normal dot layout. The dot disorder degree is normalized to the range of 0 to 1 by using the min-max normalization method. The larger the value, the more obvious the dot anomaly. The weighting coefficient for the percentage of defect area is used to adjust the contribution of the indicator to the severity of the defect. It reflects the importance of the indicator under different printing plate types and can be set according to the characteristics of the printing plate and experience with defect sensitivity. For example, the text area can increase the weight of texture disturbance, and the large color block area can increase the weight of color deviation. The range is 0 to 1, and the sum of all weights is 1. : Weighting coefficient for texture perturbation intensity; Weighting coefficient for the degree of color deviation; Weighting coefficients for the degree of branch network disorder; The severity of a defect is obtained by weighted summation of four types of defect influencing factors.

[0034] Furthermore, the percentage of the defect area includes:

[0035] in: The actual number of pixels in the defect area is obtained based on the defect identification in S2. : The total number of pixels in the corresponding detection area, obtained based on the area division results.

[0036] Furthermore, the extraction of the abnormal feature signals includes: Texture distribution anomaly:

[0037] in: The intensity of texture disturbance in a region measures the deviation of the texture in that region from the texture in a normal region. It is used to detect scratches, missing prints, or minor surface structure anomalies and to identify defective areas. : Region texture entropy value, calculated by analyzing the grayscale distribution of the image; : The mean texture entropy of the normal region is calculated by extracting texture features of the non-defect region using S1.

[0038] Color fluctuation:

[0039] in: The degree of color deviation in a region measures the overall color deviation of that region relative to the normal region. It is used to determine whether there are color-related defects in that region and to identify defective regions. : No. The color difference value of each pixel is calculated using the CIE-Lab color space and the standard sample, that is, the Euclidean distance between two pixels in the CIE-Lab color space is calculated. Total number of pixels in the region.

[0040] Changes in dot density:

[0041] in: The degree of dot misalignment in a region measures the deviation of the dot distribution in that region from the normal dot layout. It is used to detect dot anomalies caused by ink droplets, ink accumulation, or uneven printing, and to identify defective areas. The standard deviation of the network density in a region is calculated by analyzing the distribution of network points within the defective region. The standard deviation of the dot density in the normal area is calculated by collecting data from the non-defective area using S1.

[0042] S3. Perform color deviation analysis based on defect information to obtain color deviation information.

[0043] In this embodiment, the color deviation analysis includes: Most printing quality issues are essentially reflected directly in color performance. For example, insufficient ink deposition leads to increased brightness, excessive ink causes dark area accumulation, and misregistration causes color edge misalignment. Therefore, defect identification alone cannot determine the actual visibility of a quality problem. Without quantifying color deviation, it is impossible to distinguish between slight ink deviation and severe color deviation, and no quantitative basis can be provided for subsequent evaluation.

[0044] Based on the obtained defect information, the pixels of the defect area are extracted. Through local color difference calculation and aggregation, the color difference performance of the defect area is evaluated, revealing the ink color fluctuation within the area and obtaining the color deviation information of the defect area. This improves the sensitivity of local color deviation perception and enhances the identification of traditional low-sensitivity defects such as ink accumulation, slight scratches, and minor omissions. This is more in line with the needs of printing quality inspection, reduces the probability of missed detection, and improves the reliability of defect judgment.

[0045] Furthermore, the local color difference calculation and aggregation includes: Pixels in the normally printed area are selected as a control, and their color differences are compared pixel by pixel with those in the defective area. The defective area is divided into several micro-units, and the local color difference value of each micro-unit is calculated using pixel-level color difference data. The color difference data is then aggregated to obtain the regional deviation result. The specific implementation steps include: Color comparison area determination: Based on the defect information output in step S2, determine the coordinates of the defect area and locate the pixel range that needs to be analyzed for color deviation. For each defect area, select a certain range of normal printing area around it as the control area through an artificial intelligence recognition network. For example, extend a certain distance (e.g., 5 to 10 pixels) outward from the boundary of the defect area to construct the control area, so that the control area is representative and has sufficient sample density to compare the brightness and hue difference between the defect area and the normal area, reflecting the actual printing press state deviation. The size of the control area can be set according to the printing plate accuracy, printing resolution, and common dot sizes. The image data of the control area can be directly extracted from the original image collected in S1 to ensure the consistency of the data source, thereby avoiding the single sample due to the use of the defect area alone for calculation, making the deviation assessment more robust. Converting regional pixels to a unified color space: Using the existing mature RGB-Lab color mapping algorithm, the color value characteristics are ensured to be consistent with industry standards. Pixels in both defective and control areas are converted to the CIE-Lab color space. This space separates brightness (luminance), red-green channel offset, and yellow-blue channel offset, so that all subsequent color difference analysis is calculated based on a unified color language. This eliminates interference factors such as lighting environment, equipment exposure differences, and uneven channel distribution, enabling subsequent calculations to more accurately reflect issues such as ink volume deviation, overprinting, underprinting, and tone drift. Pixel color difference value calculation: For any pixel in the defect area, a reference pixel is selected from the control area for comparison using the nearest neighbor matching principle. The Lab component values ​​of the corresponding pixels in the defect area and the control area are read, and the color difference value of each pixel is calculated to obtain the initial color difference distribution reflecting the degree of color deviation. The color change trajectory is recorded in pixels. Micro-unit local deviation calculation: The defect area is divided into multiple micro-unit blocks according to a fixed number of pixels. The average color difference of each unit is calculated separately to generate a local deviation index. The dot coverage in each unit is marked to provide a basis for subsequent correlation analysis with dot anomalies. This ensures that the average value of the whole area is not diluted by the presence of severe local color deviation. At the same time, it can present details such as uneven ink layer distribution, color breakpoints, and local overlapping marks. It effectively distinguishes between two different types of uniform slight color deviation and local abrupt color deviation, making the results more targeted and interpretable. Color deviation information aggregation: All micro-unit deviations are aggregated and integrated to obtain regional color difference values. All color deviation information is summarized to form the color deviation information of the defective area and stored as structured output. The format includes pixel color difference values, micro-unit local color difference values ​​and regional color difference values ​​for each defective area.

[0046] Furthermore, the pixel color difference value calculation includes:

[0047] in: Pixel color difference value, representing the defect area's first pixel. The degree of color deviation between each pixel and its corresponding control pixel is used to quantify the color anomaly of a single pixel. The larger the value, the more obvious the deviation, providing a basis for local and overall deviation statistics. Defect area number The brightness of each pixel That is, the first comparison area The brightness of each pixel; Defect area number The red-green offset of each pixel represents the red-green channel offset of the pixel in the CIE-Lab color space. This is beneficial for eliminating lighting interference in color difference calculations and is obtained through the mature RGB-Lab color mapping algorithm. That is, the first comparison area A red-green offset of one pixel; Defect area number The yellow-blue shift of each pixel represents the pixel's shift in the yellow and blue channels of the CIE-Lab color space. This helps eliminate lighting interference in color difference calculations and is obtained through the mature RGB-Lab color mapping algorithm. That is, the first comparison area Yellow-blue offset of 1 pixel; The color difference value of each pixel is obtained by calculating the Euclidean distance between the defective region pixel and the corresponding control region pixel in Lab space.

[0048] Furthermore, the calculation of the local deviation of the micro-unit includes:

[0049] in: : Local average color difference value of micro-units, representing the defect area's first... The average color difference of each micro-unit reflects the intensity of local color deviation and is used for subsequent regional aggregation and local anomaly analysis. The larger the value, the more obvious the color deviation in the micro-unit; the smaller the value, the closer to the normal color. The number of pixels in a micro-unit is calculated by dividing the defect area into micro-units based on their side length and pixel resolution. The local average color difference value of the micro-unit is obtained by summing and averaging the color difference values ​​of all pixels within the micro-unit.

[0050] Furthermore, the aggregation of color deviation information includes:

[0051] in: The area color difference value represents the overall color deviation of the entire defect area relative to the adjacent normal area. The larger the value, the more serious the deviation. By quantifying the color deviation from the local to the overall, more precise color measurement is achieved, which can better reflect the actual state deviation of the printing press. Number of micro-units.

[0052] S4. Calculate the registration deviation based on the color deviation information to obtain the registration information.

[0053] In this embodiment, the calculation of the registration deviation includes: In color printing, it is often necessary to overlay multiple ink plates, such as C, M, Y, and K, for output. If mechanical misalignment occurs between the color plates during transmission, positioning, or tension control, registration defects such as misalignment, ghosting, and blurry edges will appear. Based on the color anomaly areas identified by S3, and avoiding noisy areas, the calculation focus is placed on the color layer overprinting area with clear and reliable edges, thereby ensuring the accuracy and repeatability of registration deviation analysis.

[0054] Using the obtained color deviation information, relatively uniform and clearly defined overlapping areas are selected as the basic areas for registration analysis. The spatial relationship between each color plate is clarified, rotation and scaling deviations are identified, and registration information is obtained. This yields a quantifiable registration deviation index, determines the printing color overlap accuracy, reflects the registration quality level, and provides a basis for quality control.

[0055] The implementation steps include: Extracting reliable analysis areas: Using the color deviation information output in step S3 as the screening basis, the artificial intelligence recognition network marks areas with significant color differences as low confidence areas according to threshold rules (which can be set according to actual needs), and removes these areas from the overall printed image. Only relatively uniform and clear overlapping color areas are retained as the basic areas for registration analysis, thereby avoiding the offset of registration positioning caused by defective color blocks and ink accumulation areas, improving the success rate of subsequent edge detection, ensuring the credibility of the entire registration calculation, and improving the consistency of feature extraction. Extracting multi-color channel feature edges: Edge detection is performed separately for independent channels such as C, M, Y, and K. The mature Canny edge detection algorithm is used to obtain the edge contour of the page. The edge data is used for registration feature analysis. The extracted edges of each color channel must be consistent with the corresponding positions of the screening results. Only edge pixels of the reliable area are retained as output to ensure the reliability of the feature source and avoid invalid noise interference. Establish a standard template color reference channel: Select one color panel from the normal area or standard sample as the reference reference channel, and map the other color channels to the same image coordinate system. This reference channel serves as the registration target, so that the subsequent matching process has a reference, ensuring that the deviation measurement output direction is clear and facilitating quantification and comparison. Printing plate registration and alignment calculation: Based on the edge features between the reference channel and the target channel, mature digital image registration technology is used to sequentially perform key point matching, edge contour alignment and geometric difference estimation, record the horizontal offset, vertical offset and overprint angle change parameters, and improve the positioning robustness by using a continuous multi-point alignment method, thereby realizing the observable quantification of registration deviation; Output registration deviation information: Statistically analyze the registration deviation and rotational offset trends between different printing color channels and the reference channel, and output the registration deviation information to determine whether the printing equipment is in a normal registration state, whether there is a serious registration imbalance, and whether the printing machinery position needs to be adjusted or the paper tension needs to be recalibrated.

[0056] Furthermore, the registration and alignment calculation includes: Overall offset degree:

[0057] in: The overall registration deviation of the printed page reflects the degree of misalignment. The larger the value, the worse the registration. It is used for the final quality score. : No. The lateral offset of each matching point is obtained by calculating the difference between the coordinates of the corresponding position of the matching point after registration and the edge coordinates of the reference color channel. : No. Vertical offset of each matching point; : The number of valid edge points that have completed the pairing; The registration deviation is calculated by combining the horizontal and vertical registration offsets of all matching points. Angular offset estimation:

[0058] in: : Overall rotational offset trend, positive values ​​indicate clockwise offset, negative values ​​indicate counterclockwise offset, and the range is [-π,π]; By combining the offset direction angles of all matching points, the overall rotational offset trend can be calculated.

[0059] in: : No. The offset angle of each matching point, with positive values ​​indicating clockwise offset and negative values ​​indicating counterclockwise offset, ranges from [-π, π], and is used to reveal ghosting and misregistration caused by printing cylinder angle errors.

[0060] S5. Conduct a comprehensive quality evaluation based on the registration information to obtain quality evaluation information.

[0061] In this embodiment, the comprehensive quality evaluation includes: Print quality is affected by a variety of factors, and relying solely on defects, color deviations, or registration errors cannot accurately reflect the overall quality. When there are slight color differences but no defects, or when there is significant registration misalignment but the overall color performance is good, independent indicators are insufficient to provide a stable judgment.

[0062] Based on the registration information, combined with defect information and color deviation information, a comprehensive quality evaluation is conducted on each printed image. The quality level is then determined based on the scoring results, providing a unified, objective, and quantifiable evaluation basis for printing quality. This helps to pinpoint the source of quality problems, further support the formulation of quality improvement strategies, improve the efficiency of automated detection and screening, and reduce human error.

[0063] The methods for comprehensive quality evaluation include: Convert local indicators into overall indicators: For the defect severity in step S2 and the color difference value of the area in step S3, each area is weighted and averaged according to area or visual weight. The local indicators are summarized into overall indicators to ensure that defects with larger areas or stronger visual impact contribute more to the overall score. This ensures that the local indicators can reasonably reflect the average situation of the entire printed page, and prepares for subsequent integration with registration information. Indicator normalization processing: The overall defect severity, overall color difference value and the registration deviation in the registration information obtained from S4 are normalized according to the set upper limit, so that they are converted into evaluation quantities with an interval of [0,1], eliminating the difference in dimensions and ensuring that indicators with different physical meanings are comparable. Weighting coefficients for each indicator: Based on the printing type, number of colors and process requirements, weights are set for the three indicators, which respectively represent the importance of the overall defect severity, overall color difference value and registration deviation on the quality. The weights can be determined by experimental statistics or industry standards, and the sum of the weights of the three indicators is 1. Comprehensive quality score calculation: The three normalized indicators are weighted and summed to obtain the overall quality score of the printed matter. The quality score ranges from 0 to 1. The larger the value, the worse the overall quality. This ensures that the contribution of each type of indicator is reflected in the final score according to the preset weight. Quality levels are determined based on the scoring results: the overall score is mapped to a quality level. For example, 0.2 or less is excellent, 0.4 or less and greater than 0.2 is acceptable, 0.7 or less and greater than 0.4 requires improvement, and greater than 0.7 is unacceptable. The threshold can be adjusted according to production standards. Output comprehensive quality evaluation information: Output the level and corresponding comprehensive score to form structured data, including the normalized value of the original indicators, weight allocation and final score, to provide traceable quantitative basis for process control, quality inspection judgment and historical data statistics.

[0064] S6. Conduct quality analysis based on quality evaluation information to obtain quality monitoring results.

[0065] In this embodiment, the quality analysis includes: In printing production, a single inspection result cannot fully guide production control. It is necessary to record and analyze the quality results of each plate to form a long-term data accumulation. Recurring defects, color deviations, or registration deviations are often related to equipment operating status, process parameters, or ink characteristics. Data analysis is needed to discover patterns and reduce the defect rate. Through recording and statistical analysis, managers can quickly understand the printing quality level of each batch and decide whether it is necessary to adjust printing equipment, recalibrate the plate, or optimize the ink ratio.

[0066] Based on the obtained quality evaluation information, the data is classified and statistically analyzed to form trend analysis data. By comparing it with historical data, recurring high-risk areas, common defect types, and abnormal production patterns are identified, enabling anomaly detection and pattern discovery. This yields quality monitoring results, providing a basis for decision-making in printing equipment maintenance, ink management, production parameter adjustment, and process improvement, and supporting production management and process optimization.

[0067] The implementation methods for this step include: Data structure design: Based on the comprehensive score and grade output by S5, each printed page information is encapsulated into a record unit, including page ID, defect location, defect severity, color deviation, registration deviation, comprehensive score and quality grade. The data structure should support batch identification, production date, printing press information and ink type to facilitate cross-batch statistical analysis. Data storage: The mature industrial information management system integration technology is used to store the record units into the database, ensuring storage security, stability and reliability, and supporting fast query and batch operation. The database can adopt hierarchical storage, such as the basic index layer (original defect and deviation data), the comprehensive scoring layer (S5 output), and the analysis results layer (statistical and trend analysis). Batch and area classification: Classify and statistically analyze by production batch, printing press, and plate area to obtain quality distribution and frequency. Statistical analysis can be performed on the proportion of defect types, average color deviation, and average registration deviation for each category, providing basic data for anomaly analysis. Anomaly pattern analysis: Using an artificial intelligence recognition network, it compares with historical data to identify recurring defect areas or common defect types, marks high-frequency anomaly areas, helps operators identify potential problems in advance, and analyzes inter-batch trends to discover the impact of changes in printing equipment or process parameters on quality. Output quality monitoring results: The analysis results will be generated into structured data, including the overall quality level distribution, local high-risk areas, quality change trends and historical comparison data, which will be available for production management personnel to view and serve as a basis for decision-making on process improvement, equipment maintenance and ink management.

[0068] Example 2: Reference Figure 2 This is the second embodiment of the present invention, which provides an artificial intelligence-based printing quality monitoring system, including the following functional modules: Feature acquisition module: responsible for acquiring images of the printed page, obtaining the original visual information of the printed page, performing preliminary image processing, and extracting initial features of the printed page, such as texture, color distribution, edge information, etc., to provide input for subsequent defect identification; Defect identification module: Uses the initial features provided by S1 to identify defect areas, including abnormal areas such as missing ink, stains, and scratches, and calculates the severity of each defect area to form defect information data; Color Analysis Module: Uses S2 defect area information to calculate the color deviation of each area, generates color deviation information, and detects problems such as uneven ink color, color deviation or color drift during the printing process; Registration Analysis Module: Based on S3 color deviation information, calculates the registration deviation between multi-color printing plates, which is used to determine the alignment accuracy of printing plates, identify misalignment or offset problems, and output registration information. Quality scoring module: It comprehensively utilizes the defect information, color deviation information and registration information output by S2, S3 and S4, and generates an overall quality score and quality grade through weighted scoring. It realizes the conversion of local indicators into overall scores, provides an operable evaluation of printed materials quality, and provides a basis for production management and quality decision-making. Quality monitoring module: Receives S5 comprehensive score and quality level information, records, classifies, statistically analyzes and anomalies to form quality monitoring results, including batch statistics, high-frequency defect areas, trend analysis and historical comparison data, providing production process monitoring, anomaly warning and process improvement reference, and realizing closed-loop quality management.

[0069] Example 3: In practical applications, this invention can be applied to high-precision color packaging printing production lines for comprehensive quality monitoring and online anomaly warning of printed materials, with the typical scenario of ensuring the accuracy of printed graphics, colors, and registration.

[0070] In this application, the printing plate is first captured using a high-resolution industrial camera to obtain a complete image of the plate. Texture, edge, and color features are extracted from the image to form initial features for subsequent analysis. Subsequently, artificial intelligence algorithms are used to identify defect areas using the initial features, detecting defects such as missing ink, scratches, and stains. Each defect area is then quantified to form quantifiable defect information.

[0071] The colors of each defective area are compared and calculated with the standard colors of the normal area to obtain local color deviation information. In the multi-color printing plate, plate registration and alignment calculations are performed to obtain registration deviation information and obtain the registration deviation amount to accurately reflect the alignment accuracy between different color layers.

[0072] Local indicators are aggregated according to area or visual weight to generate overall defect severity and overall color difference indicators. The overall defect severity, overall color difference, and registration deviation are then weighted according to preset weights to generate a comprehensive print quality score and classify quality levels, thus realizing the conversion from local indicators to overall quality evaluation.

[0073] Finally, the quality grades and related indicators of each printing plate are recorded in the database, and batch statistics, local high-risk area analysis and trend analysis are performed to form structured quality monitoring results. This supports production management personnel in making decisions on printing process adjustments, equipment maintenance and anomaly warnings, ensuring the stability of the printing production line and the consistency of print quality.

[0074] Example 4: This embodiment also provides a computer device applicable to an artificial intelligence-based printing quality monitoring method and system, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the artificial intelligence-based printing quality monitoring method and system proposed in the above embodiment.

[0075] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements an artificial intelligence-based printing quality monitoring method and system as described in the above embodiments.

[0076] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0077] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0078] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0079] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0080] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0081] 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, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A printing quality monitoring method based on artificial intelligence, characterized in that, Includes the following steps: S1. Collect the printing plate layout to obtain initial features; S2. Based on the initial features, identify the defect area and obtain defect information; S3. Perform color deviation analysis based on defect information to obtain color deviation information; S4. Calculate the registration deviation based on the color deviation information to obtain the registration information; S5. Conduct a comprehensive quality evaluation based on the registration information to obtain quality evaluation information; S6. Conduct quality analysis based on quality evaluation information to obtain quality monitoring results.

2. The printing quality monitoring method based on artificial intelligence according to claim 1, characterized in that, The printing plate acquisition in step S1 includes: By continuously acquiring images of the printed page using acquisition tools, complete and accurate information about the printed page is obtained, and initial features are extracted from this information.

3. The printing quality monitoring method based on artificial intelligence according to claim 1, characterized in that, The defect region identification in step S2 includes: Based on the initial features obtained, defect classification and identification and defect severity quantification are achieved through region division and abnormal feature signal extraction, thereby obtaining defect information, achieving accurate defect location, and providing statistical indicators for the final quality score.

4. The printing quality monitoring method based on artificial intelligence according to claim 3, characterized in that, The quantification of the degree of defect includes: ； in: Defect severity; : Percentage of defective area; Normalized texture perturbation intensity; : The degree of color deviation during normalization; The degree of disorder in the normalized network nodes; : Weighting coefficient for the percentage of defective area; : Weighting coefficient for texture perturbation intensity; : Weighting coefficient for the degree of color deviation; Weighting coefficients for the degree of branch network disorder; The severity of a defect is obtained by weighted summation of four types of defect influencing factors.

5. The printing quality monitoring method based on artificial intelligence according to claim 1, characterized in that, The color deviation analysis in step S3 includes: Based on the obtained defect information, the pixels in the defect area are extracted. Through local color difference calculation and aggregation, the color difference performance of the defect area is evaluated, the ink color fluctuation within the area is revealed, and the color deviation information of the defect area is obtained.

6. The printing quality monitoring method based on artificial intelligence according to claim 5, characterized in that, The local color difference calculation and aggregation includes: Pixels in the normal printing area are selected as a reference, and the color difference is compared pixel by pixel with those in the defective area. The defective area is divided into several micro-units, and the local color difference value of each micro-unit is calculated using pixel-level color difference data. The color difference data is then aggregated to obtain the regional color deviation.

7. The printing quality monitoring method based on artificial intelligence according to claim 1, characterized in that, The calculation of the registration deviation in step S4 includes: Using the obtained color deviation information, relatively uniform and clearly defined overlapping color areas are selected as the base area for registration analysis. The spatial relationship between each color plate is clarified, rotation and scaling deviations are identified, and registration information is obtained.

8. The printing quality monitoring method based on artificial intelligence according to claim 1, characterized in that, The comprehensive quality evaluation in step S5 includes: Based on the registration information, combined with defect information and color deviation information, a comprehensive quality evaluation is conducted on each printed image, and the quality level is divided according to the scoring results to obtain quality evaluation information.

9. The printing quality monitoring method based on artificial intelligence according to claim 1, characterized in that, The quality analysis in step S6 includes: Based on the obtained quality evaluation information, the data is classified and statistically analyzed to form trend analysis data. By comparing it with historical data, recurring high-risk areas, common defect types, and abnormal production patterns are identified, thereby achieving anomaly detection and pattern discovery and obtaining quality monitoring results.

10. An artificial intelligence-based printing quality monitoring system, constructed based on the artificial intelligence-based printing quality monitoring method according to any one of claims 1-9, characterized in that, Includes the following functional modules: The module includes a feature acquisition module, a defect identification module, a color analysis module, a registration analysis module, a quality scoring module, and a quality monitoring module.

Images