Printing machine real-time monitoring method based on image processing and related equipment
By implementing a real-time monitoring method based on image processing on the printing press, using image acquisition equipment and machine learning algorithms, the problem that the existing technology cannot monitor the quality of printed materials in real time is solved, timely detection and correction of printed materials are realized, and the quality of printed materials is improved. Productivity.
Patent Information
- Application Number
- CN202411279099.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-12
- Publication Date
- 2024-11-08
AI Technical Summary
Existing printing press monitoring technology cannot monitor the printing quality of printed materials in real time, and cannot promptly discover and deal with possible quality problems during the printing process, such as color deviations, printing errors, etc.
Using an image processing-based method, by setting up multiple types of image acquisition devices on the printing press, we can obtain the static images and dynamic video streams of printed materials, and use image enhancement technology, color space conversion, object detection algorithms and machine learning algorithms to monitor and evaluate printed materials in real time. The quality of the test box is dynamically adjusted, and information is transmitted with the printing press control system through the real-time communication link, and parameters are automatically adjusted to correct quality problems.
Real-time monitoring and evaluation of the quality of printed materials is realized, the defective rate is reduced, the overall quality of printed materials is improved, manual inspection errors and labor intensity are reduced, production interruptions and resource waste are reduced, and the stable operation and production of printing presses are ensured. Continuity.
Smart Images

Figure CN118906649A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of printing technology and image processing technology, and in particular to a real-time monitoring method for a printing press based on image processing and related equipment. Background Art
[0002] At present, the monitoring technology of printing press has been widely used, but it mainly focuses on monitoring the running status of printing machine parts and the consumption of printing materials. Although these monitoring methods can provide certain production data, they cannot directly reflect the quality of printed products.
[0003] The existing printing press monitoring system mainly collects the operating data of mechanical parts, such as temperature, pressure, speed, etc., through sensors, and makes judgments based on preset thresholds to realize the status monitoring of the printing press. The existing technology cannot monitor the printing quality of printed products in real time, and cannot timely discover and deal with quality problems that may occur in the printing process, such as color deviation and printing errors.
[0004] Therefore, based on the above problems, the prior art still needs to be improved. Summary of the invention
[0005] The purpose of this application is to provide a real-time monitoring method and related equipment for a printing press based on image processing, aiming to solve the problem that the printing quality of printed products cannot be monitored in real time and that quality problems that may occur during the printing process cannot be discovered and handled in a timely manner.
[0006] The first object of this application is to provide a real-time monitoring method for a printing press based on image processing, comprising: On a high-speed printing production line, the printer uses multiple types of image acquisition devices to obtain static images and dynamic video streams of printed products at different angles and lighting conditions at each printing stage. Based on the static image and the dynamic video stream, screening is performed to obtain a captured image; For complex color prints, specific image enhancement technology and color space conversion technology are used to respectively enhance the clarity and contrast of the captured image, and convert the captured image into a corresponding color space; Using the target detection algorithm in deep learning, the elements on the printed matter are preliminarily identified and located, and then the detection frame is adaptively divided according to the characteristics of the elements, and the position and size of the detection frame are dynamically adjusted in real time; For the preprocessed collected image, multiple edge detection algorithms are used to obtain multiple sets of edge detection results accordingly, and relevant edge information is obtained through a fusion strategy; The optimization algorithm in machine learning is used to calculate the optimal grayscale threshold range, and the printing quality of the printed product under different ambient lights is evaluated based on the optimal grayscale threshold range to obtain an evaluation result.
[0007] By adopting the above technical solutions, through real-time monitoring and multi-dimensional image acquisition and analysis, it is possible to timely discover and correct quality problems in the printing process, reduce the defective rate, and improve the overall quality of printed products. Real-time dynamic adjustment of the detection frame and rapid evaluation of printing quality can help reduce pauses and adjustment time in production and achieve efficient operation of high-speed printing production lines. Special processing technology for complex color prints can meet diversified and high-quality printing requirements and expand the business scope. With the help of deep learning and machine learning algorithms, automated quality inspection and evaluation are achieved, reducing the error and labor intensity of manual inspection. Based on the evaluation results of printing quality, data support can be provided for the optimization of printing technology, continuously improving production processes and improving production efficiency.
[0008] In a possible implementation of the present application, the method further includes: Build a real-time communication link between the detection system and the printing press control system; When it is detected that the printing quality of the printed product exceeds a preset threshold, in a large-scale printing production scenario, an early warning signal is sent and relevant parameters of the printing press are automatically adjusted.
[0009] By adopting the above technical solutions, establishing a real-time communication link can ensure instant information transmission between the detection system and the printing press control system. Once a quality problem is found, it can respond quickly to avoid the expansion of the problem and minimize the production of defective products. Automatically adjusting parameters when the printing quality exceeds the threshold helps to maintain the stable operation of the printing press, reduce production interruptions caused by quality fluctuations, and ensure the continuity and stability of large-scale printing production. Automatically sending early warning signals and adjusting parameters reduces the need for manual monitoring and intervention, reduces labor costs, and avoids quality problems caused by human negligence or misjudgment. Timely adjustments can reduce rework and equipment maintenance due to quality problems, increase the effective working time of the printing press, and thus improve overall production efficiency. Automatically adjusting parameters helps to maintain the consistency of printed quality, meet customers' high requirements for product quality stability, and enhance customer satisfaction and trust. Through timely early warning and adjustment, the waste of raw materials, energy and other resources caused by the production of a large number of unqualified products is avoided. Relevant data recording and analysis can provide a basis for subsequent production optimization, help enterprises continuously improve processes and parameter settings, and achieve more efficient and higher-quality production.
[0010] In a possible implementation manner of the present application, the method further includes: Create a print quality database to store each captured image, test results, and relevant parameters for printing press adjustment; Conduct regular analysis on the data in the printed product quality database, extract trends and rules of quality changes, and obtain analysis results; Based on the analysis results, identifying key factors affecting the quality of printed products; When the captured image is acquired, a printing adjustment plan is automatically generated based on the key factors.
[0011] By adopting the above technical solutions, the creation of a quality database can fully save the images, test results and related parameters of each acquisition, facilitate the traceability and review of the production process, and help find the root cause of the problem. By regularly analyzing the data and extracting the trends and laws of quality changes, the occurrence of quality problems can be predicted in advance, preventive measures can be taken, and quality risks can be reduced. Based on the analysis results, the key influencing factors can be identified, and the root causes of quality problems can be found more accurately, so as to make targeted improvements and optimizations. When acquiring new acquisition images, the printing adjustment plan is automatically generated based on the key factors, which realizes fast and accurate production adjustments and reduces the time and errors of manual decision-making. By continuously accumulating and analyzing data, the printing process can be continuously optimized and the quality of printed products can be steadily improved. Accurate adjustment plans help reduce material waste, equipment loss and energy consumption caused by quality problems, and reduce production costs. The automatically generated adjustment plan can be quickly adapted according to different production conditions and requirements, enhancing the ability of enterprises to respond to diversified orders.
[0012] In a possible implementation manner, the present application includes the following steps before applying specific image enhancement technology and color space conversion technology to complex color prints: Acquire images of prints of varying types, colors, and pattern complexity; Annotating the printed matter image to obtain annotation data, wherein the annotation data includes information such as the type, color characteristics, and pattern complexity of the printed matter; Build a multi-layer convolutional neural network model, set the convolution layer, pooling layer and fully connected layer, and determine the network structure and parameters; Using the labeled data to train the multi-layer convolutional neural network model, and continuously adjusting the weights and biases of the multi-layer convolutional neural network model through a back propagation algorithm to optimize model performance; Use the validation set to evaluate the trained multi-layer convolutional neural network model, calculate the accuracy, recall and other indicators, and adjust the structure or parameters if the requirements are not met until the evaluation result of the multi-layer convolutional neural network model is good; The collected printed images are input into the trained multi-layer convolutional neural network model to obtain the image classification results, and the printed images are classified according to the type, color, pattern complexity and other attributes of the printed materials.
[0013] By adopting the above technical solutions, by acquiring rich and diverse printed images and carefully annotating them, a high-quality data foundation is provided for subsequent model training, which helps to improve the model's processing accuracy for complex color prints. Building a multi-layer convolutional neural network model and reasonably setting each layer can effectively extract the features of printed images, thereby improving the performance and generalization ability of the model. Using annotated data for training and adjusting weights and biases through the back propagation algorithm, the model can better learn the characteristic laws of printed images and continuously optimize model performance. Using the validation set for evaluation, it is ensured that the model's accuracy and recall rate and other indicators reach a good level, thereby providing reliable support for subsequent image processing. The collected printed images are input into the trained model for classification and categorization, which can quickly and accurately organize the printed images according to relevant attributes and improve work efficiency. Good classification and categorization results provide an accurate basis for the subsequent use of specific image enhancement technology and color space conversion technology, so that these technologies can be applied more targetedly and improve the overall processing effect. It can process printed images of different types, colors and pattern complexities to meet the market's processing needs for diversified printed products. The automated classification and categorization process reduces the workload of manual processing and saves a lot of manpower and time costs.
[0014] In a possible implementation manner of the present application, when it is detected that the printing quality of the printed product exceeds a preset threshold, the following steps are included: According to the quality standards and production requirements of printed materials, pre-set the threshold of printing quality, including the upper and lower limits of various key indicators; Continuously monitor the printing quality of printed products in real time, compare the detected quality data with a preset threshold, and trigger an early warning mechanism when the quality data exceeds the preset threshold; When the warning is triggered, a command is sent to suspend the operation of the printing press; If the printing task is urgent, the system automatically switches to the pre-set backup printing mode, which uses the preset printing parameters; The products that have been printed but do not meet the quality standards are automatically marked, and the unqualified products are separated from the qualified products through mechanical devices or transmission systems for subsequent separate processing.
[0015] By adopting the above technical solutions, the thresholds of printing quality are pre-set, including the upper and lower limits of various key indicators, so that the quality standards are clear and provide an accurate measurement benchmark for the production process. Continuously monitoring the printing quality in real time and comparing it with the preset threshold can timely detect the situation where the quality exceeds the threshold and avoid further deterioration of the problem. Triggering an early warning and suspending the operation of the printing press can effectively prevent the production of more unqualified products, reduce the defective rate, and save costs. When the printing task is urgent, switch to the backup printing mode to ensure the continuity of the production task and reduce production delays caused by quality problems. Automatically mark and isolate unqualified products for subsequent separate processing, which improves the efficiency and accuracy of quality management. Avoiding the subsequent processing and handling of unqualified products saves manpower, material and time resources. By timely discovering and handling quality problems, it helps to continuously improve the production process and improve the overall quality level of printed products. The entire process makes the printing production process more controllable and improves the company's ability to control production quality and respond to emergencies.
[0016] In a possible implementation of the present application, the method includes: Collect various data from the print quality database, including image data, test results, printing machine parameters, environmental data, etc., and integrate and standardize them; Eliminate noise, missing values and outliers in the data to ensure the accuracy and completeness of the data; Extracting features from the data, including but not limited to color distribution features of statistical images and change trend features of printing press parameters; Select a preset big data analysis model and apply it to data mining tasks to obtain mining results; Analyze and interpret the mining results to derive potential quality problem patterns, key factors related to quality, and possible improvement directions; The mining results are visualized.
[0017] By adopting the above technical solutions, various types of relevant data are collected from the quality database and integrated and standardized, breaking the data island and making the data more comprehensive and usable. Noise, missing values and outliers are eliminated to ensure that the data used for analysis is accurate and complete, and the reliability of the analysis results is improved. Extracting rich features from the data helps to discover potential patterns and associations, providing a deeper perspective for solving quality problems. Selecting a suitable big data analysis model for data mining can more effectively mine valuable information and hidden patterns. Analyzing and interpreting the mining results, clarifying the potential quality problem patterns and key factors, and making problem solving more targeted. Possible improvement directions are obtained to provide a scientific basis for optimizing printing processes and quality control strategies and promote continuous improvement. Visualizing the mining results makes complex data easier to understand and interpret, providing decision makers with intuitive and clear information and accelerating the decision-making process.
[0018] In a possible implementation of the present application, the method includes: Based on the analysis results of the printed product quality database and the identification of key factors, the adjustment plan of the printing press parameters is automatically generated using machine learning algorithms; Pushing the adjustment plan to the terminal of the printing press operator in real time through a human-computer interaction interface or a mobile terminal application, while providing a detailed description of the plan and the expected effect; Obtain the corresponding parameter adjustments and operations performed by the operator on the printing press control system; The system automatically collects actual printing quality data and equipment operating status data after the operator implements the adjustment plan; Based on the actual printing quality data and equipment operation status data, combined with the new detection results of the collected images, the original adjustment plan is optimized and improved to improve the accuracy and effectiveness of the plan; With the continuous advancement of the production process and the continuous accumulation of data, the quality feedback mechanism is continuously updated and improved to adapt to different production conditions and quality requirements.
[0019] By adopting the above technical solutions, the adjustment plan is automatically generated based on database analysis and key factor identification, which improves the efficiency and scientificity of plan formulation and reduces the subjectivity and errors of manual decision-making. The adjustment plan and instructions are pushed in real time through the human-computer interaction interface or mobile terminal application to ensure that operators can obtain the latest and most accurate guidance in a timely manner and reduce information transmission delays. Operators adjust parameters and operate according to the pushed plan, which ensures the standardization and consistency of operation and helps to improve the stability of printing quality. The system automatically collects relevant data after implementation, providing a real-time and accurate basis for the optimization of the plan, so that the adjustment plan can continuously improve itself. The adjustment plan is optimized and improved in combination with the new test results, which significantly improves the accuracy and effectiveness of the plan and better meets production needs. The quality feedback mechanism is continuously updated and improved so that it can adapt to different production conditions and quality requirements, enhancing the adaptability and robustness of the entire system. By continuously optimizing the adjustment plan and feedback mechanism, the continuous improvement of printing quality is promoted and the competitiveness of products is improved.
[0020] The second object of the present application is to provide a real-time monitoring system for a printing press based on image processing, the system comprising: Static image and dynamic video stream acquisition module: used on a high-speed printing production line, through the multi-type image acquisition devices set up by the printing press, to obtain static images and dynamic video streams of the printed products at different angles and light conditions for the printed products at various printing stages; Static image and dynamic video stream screening module: based on the static image and the dynamic video stream, screening is performed to obtain the captured image; Captured image preprocessing module: for complex color prints, specific image enhancement technology and color space conversion technology are used to respectively improve the clarity and contrast of the captured image, and convert the captured image to the corresponding color space; Detection frame division and adjustment module: using the target detection algorithm in deep learning to preliminarily identify and locate the elements on the printed matter, then adaptively dividing the detection frame according to the characteristics of the elements, and dynamically adjusting the position and size of the detection frame in real time; Related edge information acquisition module: for the pre-processed collected image, multiple edge detection algorithms are used to obtain multiple sets of edge detection results accordingly, and related edge information is obtained through a fusion strategy; Printing quality assessment module: uses the optimization algorithm in machine learning to calculate the optimal grayscale threshold range, and evaluates the printing quality of the printed product under different ambient lights based on the optimal grayscale threshold range to obtain an evaluation result.
[0021] By adopting the above technical solutions and working together with multiple modules, comprehensive monitoring of printed products at different stages, angles and lighting conditions is achieved, ensuring the accuracy and comprehensiveness of printing quality evaluation. The acquisition image preprocessing module can significantly improve the image quality of complex color printed products, provide a clearer and more accurate basis for subsequent detection and evaluation, and reduce misjudgment. The detection frame division and adjustment module can accurately identify and locate elements on printed products, improving the accuracy and efficiency of detection. The relevant edge information acquisition module uses a variety of edge detection algorithms and fusion strategies to obtain more comprehensive and accurate edge information, which helps to find subtle quality problems. The printing quality evaluation module uses an optimization algorithm to calculate the optimal grayscale threshold range, making quality evaluation more scientific and objective, and providing a strong basis for quality control. Real-time monitoring and evaluation can timely discover quality problems in the printing process, facilitate rapid measures to adjust and improve, and reduce the production of defective products. The efficient and accurate monitoring system reduces production interruptions and rework caused by quality problems, and improves the overall operation efficiency of the printing production line. Based on the evaluation results, the printing process parameters can be optimized in a targeted manner to improve the stability and consistency of the printing quality. Improvements in reducing defective products, improving production efficiency, and optimizing process parameters can help reduce production costs and improve the economic benefits of the enterprise.
[0022] The third object of the present application is to provide a real-time monitoring device for a printing press based on image processing, the device comprising: A memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and execute the above-mentioned real-time monitoring method for a printing press based on image processing.
[0023] The fourth objective of this application is to provide a storage medium.
[0024] The fourth application objective of this application is achieved through the following technical solutions: A storage medium stores a computer program that can be loaded by a processor and execute the above-mentioned real-time monitoring method for a printing press based on image processing.
[0025] In summary, the present application includes at least one of the following beneficial technical effects: 1. It realizes real-time monitoring and quality assessment of the entire printing production process, significantly improving the stability and consistency of printed quality. It greatly reduces the workload and errors of manual inspection, reduces labor costs, and improves inspection efficiency and accuracy. It can timely discover and solve quality problems in the printing process, effectively reduce the production of defective products, and reduce production costs and resource waste. Based on the application of big data and machine learning algorithms, it provides a scientific basis for the optimization and improvement of printing technology, and promotes technological innovation and industrial upgrading. The adaptive detection and adjustment functions enable the system to flexibly respond to printing tasks of different types and complexities and changes in production environment. The perfect quality feedback mechanism helps enterprises accumulate production experience and quality data, and provides strong support for long-term development and market competition. The intelligent printing press monitoring system has improved the production level and automation level of the entire printing industry and promoted the intelligent development of the industry. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 It is a flow chart of a method for real-time monitoring of a printing machine based on image processing provided in an embodiment of the present application; Figure 2 It is a virtual structural diagram of a printing press real-time monitoring system based on image processing provided in an embodiment of the present application. DETAILED DESCRIPTION
[0027] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0028] In addition, the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article, unless otherwise specified, generally means that the associated objects before and after are in an "or" relationship.
[0029] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.
[0030] The present application embodiment provides a printing press real-time monitoring method based on image processing, referring to Figure 1 , the main process of the method is described as follows: S1: On a high-speed printing production line, the printer uses a variety of image acquisition devices to obtain static images and dynamic video streams of printed products at different angles and lighting conditions at various printing stages. Among them, various types of image acquisition equipment, including high-definition industrial cameras and high-speed cameras, are installed on the high-speed printing production line. During the plate-making stage of the printed product, the high-definition industrial camera installed above collects static images of the printed product at a vertical angle, focusing on capturing the pattern and text details of the layout. At the same time, the high-speed camera installed on the side collects dynamic video streams at a 45-degree angle to record the color changes and ink flow of the printed product during the plate-making process. When entering the printing stage, image acquisition devices distributed in different positions work simultaneously. The camera located at the feed inlet of the printing press obtains a static image of the printed product from the front to check the initial printing effect. The multiple high-speed cameras distributed along the production line collect dynamic video streams from different angles, such as horizontal and tilted 30 degrees, to comprehensively monitor the color uniformity, pattern alignment, and possible ink splashing during the printing process. In the drying and finishing stage of the later printing, the industrial camera on the top collects static images again to evaluate whether the color saturation and pattern clarity of the printed product after drying meet the standards. At the same time, the high-speed camera on the side continuously records the dynamic video stream to observe whether there are scratches or other damage during the finishing process. The static images and dynamic video streams acquired by these multi-type image acquisition devices under different angles and lighting conditions provide rich and comprehensive data support for subsequent printing quality analysis and monitoring.
[0031] S2: Screening based on the static image and the dynamic video stream to obtain a captured image; Among them, after obtaining static images and dynamic video streams, a series of screening criteria are first set. For example, for static images, images with blurred, dark or bright light that cause loss of details are removed. For dynamic video streams, key frames are extracted according to the preset frame rate, and frames with large-area occlusion or shooting angles that seriously deviate from the preset requirements are excluded. Automatic screening is performed through specially developed image screening software. The software uses image recognition algorithms to analyze the clarity, contrast, color balance and other parameters of each static image and each extracted key frame. If the clarity of a static image is lower than the set threshold, or the color balance of a key frame seriously deviates from the normal range, it will be excluded. At the same time, manual sampling is also used as a supplementary means. The staff randomly selects a certain proportion of images and key frames that have been automatically screened for review to ensure the accuracy of the screening. For example, in a printing process, 1,000 static images and 5 dynamic video streams were obtained. After automatic screening and manual sampling, 800 static images were finally selected and 100 key frames were extracted from the dynamic video stream as acquisition images for subsequent processing and analysis.
[0032] S3: For complex color prints, using specific image enhancement technology and color space conversion technology to respectively enhance the clarity and contrast of the collected image, and convert the collected image into a corresponding color space; Among them, after obtaining the collected images, analysis found that the color levels of some pages were not clear enough and the details were blurred. In view of this situation, the image enhancement technology of histogram equalization was used to adjust the grayscale distribution of the image. By expanding the dynamic range of pixel values in the image, the details of the originally darker or brighter areas can be revealed, thereby significantly improving the clarity of the image. At the same time, for the parts with low color contrast, sharpening processing based on the Laplace operator was used. The edges and transition areas of the colors in the image are enhanced, making the boundaries between different colors clearer, greatly improving the color contrast. Then, in order to more accurately analyze and process color information, the collected images are converted from the original RGB color space to the CMYK color space that is more suitable for printed product analysis. In this conversion process, the characteristics and color expression capabilities of printing inks are fully considered. For example, when processing a page with bright flowers as the theme, after the above processing, the originally slightly dim petal color becomes more vivid, the texture of the petal edge becomes clearer, and the transition between colors becomes more natural and distinct. The visual effect of the entire page has been significantly improved, which is more in line with the requirements of high-quality printing.
[0033] S4: using a target detection algorithm in deep learning to preliminarily identify and locate the elements on the printed matter, then adaptively dividing the detection frame according to the features of the elements, and dynamically adjusting the position and size of the detection frame in real time; Consider a printed product with multiple product labels. These labels contain elements such as text, graphics, and barcodes. First, the YOLO target detection algorithm based on deep learning is used to preliminarily identify and locate the elements on the printed product. The algorithm quickly identifies the text area, graphic area, and barcode area. For the text area, a rectangular detection frame is adaptively divided according to its font size, arrangement, and other features. For example, for smaller fonts, the detection frame will be relatively reduced to more accurately frame the text. During the printing process, the position of the element may change due to slight movement or stretching of the paper. At this time, the system monitors these changes in real time and dynamically adjusts the position of the detection frame. For example, when a text area moves a little to the left due to a slight offset of the paper, the detection frame will also move to the left accordingly to ensure that the text is always accurately framed. At the same time, the size of the detection frame is adjusted in real time according to the scaling of the elements on the printed product. For example, when the graphic is enlarged or reduced, the detection frame will also be enlarged or reduced proportionally to keep it close to the edge of the graphic. Through such real-time dynamic adjustment, the accurate identification and positioning of the elements on the printed product are ensured, and the accuracy and reliability of printing quality detection are improved.
[0034] S5: for the preprocessed collected image, using a variety of edge detection algorithms, correspondingly obtaining a plurality of sets of edge detection results, and obtaining relevant edge information through a fusion strategy; Among them, the pre-processed collected image is first processed by the Sobel edge detection algorithm. The Sobel algorithm can quickly detect the edges in the horizontal and vertical directions of the image, and obtain a set of edge detection results, highlighting the more obvious straight edge parts in the pattern. Then, the image is processed by the Canny edge detection algorithm. The Canny algorithm has a good ability to suppress noise and can detect more delicate and accurate edges. Another set of edge detection results is obtained, which supplements some subtle edges and curve parts in the pattern. Then, the image also uses the Laplacian edge detection algorithm. This algorithm is sensitive to the second-order derivatives in the image and further detects some edge areas with high contrast. Through these three different edge detection algorithms, we obtain multiple sets of edge detection results accordingly. Next, a weighted average fusion strategy is used to integrate these results. For the clear and consistent edge parts in the image, a higher weight is given; for the edge parts with differences or uncertainties, a lower weight is given. For example, in the pattern of textiles, the outline of the flower is relatively clear and consistent in the detection results of different algorithms, so this part of the edge will be given a higher weight when fusion to ensure its accuracy. For some complex texture areas, the detection results of different algorithms may be different, and these edges will be given appropriate weights according to their performance in different algorithms. Ultimately, the fusion strategy is used to obtain relevant edge information that combines the advantages of multiple algorithms, more accurately depicting the edge features of textile patterns, and providing a more reliable basis for subsequent quality assessment and analysis.
[0035] S6: Calculate the optimal grayscale threshold range using the optimization algorithm in machine learning, and evaluate the printing quality of the printed product under different ambient lights based on the optimal grayscale threshold range to obtain an evaluation result.
[0036] Among them, it is assumed that a batch of printed posters need to be evaluated for quality. First, a large number of printed poster images taken under different ambient light (such as strong light, weak light, natural light, etc.) are collected as data sets. The particle swarm optimization algorithm (PSO) in machine learning is used to calculate the optimal grayscale threshold range. In the algorithm, the grayscale threshold range is used as the position of the particle, and the optimal threshold range is found according to the fitness function (for example, based on indicators such as image clarity, contrast, and color accuracy) through continuous iteration. After obtaining the optimal grayscale threshold range, the newly collected printed poster images under different ambient light are grayed. Then, the grayscale image is binarized and segmented according to the calculated optimal grayscale threshold range. By analyzing the distribution, proportion, and continuity of black and white areas in the binary image, the quality of the printed product in terms of ink uniformity, pattern clarity, and color accuracy is evaluated. For example, if a poster taken under strong light is processed and analyzed, if the black areas in the binary image are evenly distributed, the white areas are small, and the boundaries are clear, it means that the printing quality of the poster under strong light is good, the ink color is uniform, and the pattern is clear. After evaluating multiple posters under different ambient lighting conditions, comprehensive evaluation results were obtained, such as the overall printing quality is excellent, but the color accuracy of some posters under low light needs to be improved.
[0037] Specifically, in some possible embodiments, the method also includes: Build a real-time communication link between the detection system and the printing press control system; When it is detected that the printing quality of the printed product exceeds a preset threshold, in a large-scale printing production scenario, an early warning signal is sent and relevant parameters of the printing press are automatically adjusted.
[0038] In order to achieve efficient printing quality control, a real-time communication link between the detection system and the printing press control system was established. The detection system is equipped with high-precision image acquisition equipment and advanced image processing software. These devices and software can obtain images of the products being printed in real time and quickly analyze their quality parameters, such as color deviation, clarity, overprint accuracy, etc. The printing press control system is responsible for adjusting various operating parameters of the printing press, such as printing speed, pressure, ink supply, etc. When the detection system detects that the color deviation of a batch of printed products exceeds the preset threshold, it immediately sends an early warning signal to the printing press control system through a real-time communication link. After receiving the signal, the printing press control system automatically adjusts the relevant parameters. For example, reduce the ink supply to correct the problem of too dark color, or adjust the printing pressure to improve clarity. During a large-scale production of color magazines, the detection system found that the overall printing color of a page was lighter. The early warning signal was quickly sent to the printing press control system, and the system automatically increased the ink supply and fine-tuned the printing pressure, so that the color of the subsequent printed pages returned to normal, avoiding the production of a large number of defective products. By building such a real-time communication link and a mechanism for automatically adjusting parameters, the quality stability of printed products has been greatly improved, the losses caused by quality problems have been reduced, and the efficient implementation of large-scale printing production has been ensured.
[0039] Specifically, in some possible embodiments, the method further includes: Create a print quality database to store each captured image, test results, and relevant parameters for printing press adjustment; Conduct regular analysis on the data in the printed product quality database, extract trends and rules of quality changes, and obtain analysis results; Based on the analysis results, identifying key factors affecting the quality of printed products; When the captured image is acquired, a printing adjustment plan is automatically generated based on the key factors.
[0040] First, a print quality database was created. Images captured during each printing process, whether static images or key frames in dynamic video streams, as well as the corresponding detection results (such as color deviation values, clarity scores, overprint accuracy, etc.), and related parameters adjusted by the printer to ensure quality (such as pressure, speed, temperature, etc.) are stored in detail in this database. Every other week, the system will conduct regular analysis of the data in the print quality database for the past week. Data mining and statistical analysis tools are used to process a large amount of data. For example, a line graph is drawn to observe the trend of color deviation values in different time periods, or correlation is calculated to find out which parameter changes are closely related to the fluctuation of clarity scores. After analysis, it is found that the fluctuation of printing pressure has a greater impact on color uniformity, while the change of printing speed mainly affects the clarity of the pattern. Based on these analysis results, printing pressure and printing speed are determined to be the key factors affecting the quality of printed products. When a new batch of printing tasks begins and the captured images are obtained, the system will immediately automatically generate a printing adjustment plan based on the key factors previously determined. If the patterns of this batch of printed materials are more complex and require high clarity, the system will recommend reducing the printing speed appropriately; if the colors are required to be uniform and bright, it will recommend maintaining a stable and moderate printing pressure. For example, when printing a batch of color brochures, the initial images collected showed slightly uneven colors. Based on the analysis of key factors, the system automatically generated an adjustment plan to fine-tune the printing pressure by 5%. After the adjustment, the color uniformity of the subsequent printed brochures was significantly improved, reaching high-quality printing standards. In this way, data is continuously accumulated and analyzed, and the analysis results are used to optimize the printing adjustment plan, effectively improving the quality stability and production efficiency of printed products.
[0041] Specifically, in some possible embodiments, before applying specific image enhancement technology and color space conversion technology to complex color prints, the following steps are included: Acquire images of prints of varying types, colors, and pattern complexity; Annotating the printed matter image to obtain annotation data, wherein the annotation data includes information such as the type, color characteristics, and pattern complexity of the printed matter; Build a multi-layer convolutional neural network model, set the convolution layer, pooling layer and fully connected layer, and determine the network structure and parameters; Using the labeled data to train the multi-layer convolutional neural network model, and continuously adjusting the weights and biases of the multi-layer convolutional neural network model through a back propagation algorithm to optimize model performance; Use the validation set to evaluate the trained multi-layer convolutional neural network model, calculate the accuracy, recall and other indicators, and adjust the structure or parameters if the requirements are not met until the evaluation result of the multi-layer convolutional neural network model is good; The collected printed images are input into the trained multi-layer convolutional neural network model to obtain the image classification results, and the printed images are classified according to the type, color, pattern complexity and other attributes of the printed materials.
[0042] First, a large number of different types of printed images, including posters, book covers, wrapping paper, etc., were collected from multiple printing factories. These images cover a variety of colors, from single tones to rich multi-color combinations, and different pattern complexities, from simple geometric figures to complex art paintings. For each collected printed image, the system image recognition is detailedly annotated. The annotation content includes information such as the type of printed matter (such as posters, book covers, etc.), color features (such as main colors, color matching, etc.), and pattern complexity (simple, medium, complex). Then, a multi-layer convolutional neural network model was built. Multiple convolutional layers were set up to extract image features, pooling layers were used to reduce feature dimensions, and fully connected layers were used for final classification. The network structure and initial parameters, such as convolution kernel size, step size, number of layers, etc., were determined. The model was trained using the annotated data. During the training process, the weights and biases of the model were continuously adjusted through the back propagation algorithm. For example, when the model misclassifies a complex color poster, the back propagation algorithm will adjust the model parameters based on the error to improve the accuracy of the next prediction. After training, the model was evaluated using an independent validation set. Calculate indicators such as accuracy (the ratio of correctly classified images to the total number of images) and recall (the ratio of correctly classified positive examples to the actual number of positive examples). If the accuracy and recall do not meet the expected requirements (for example, the accuracy is less than 90%), adjust the network structure (increase or decrease the number of layers) or modify the parameters (such as adjusting the learning rate), and then retrain and evaluate until the model evaluation results are good. When a new batch of printed images are collected, they are input into the trained multi-layer convolutional neural network model. The model quickly gives classification results, such as classifying a colorful and complex patterned book cover image into the "complex color high complexity pattern" category. Through this implementation process, complex color printed images can be effectively and accurately classified and categorized, providing an important basis for subsequent image processing and quality assessment.
[0043] Specifically, in some possible embodiments, when it is detected that the printing quality of the printed product exceeds a preset threshold, the following steps are included: According to the quality standards and production requirements of printed materials, pre-set the threshold of printing quality, including the upper and lower limits of various key indicators; Continuously monitor the printing quality of printed products in real time, compare the detected quality data with a preset threshold, and trigger an early warning mechanism when the quality data exceeds the preset threshold; When the warning is triggered, a command is sent to suspend the operation of the printing press; If the printing task is urgent, the system automatically switches to the pre-set backup printing mode, which uses the preset printing parameters; The products that have been printed but do not meet the quality standards are automatically marked, and the unqualified products are separated from the qualified products through mechanical devices or transmission systems for subsequent separate processing.
[0044] Among them, a batch of high-end cosmetic packaging boxes were produced in a printing and packaging factory. First, according to the strict quality standards and production requirements of the cosmetics brand for packaging boxes, the threshold of printing quality was pre-set. Key indicators include the upper limit of color difference of ΔE<2, the lower limit of pattern clarity of resolution of 300dpi, and the upper limit of error of overprint accuracy of 0.1mm. During the printing process, the quality of the printed products is continuously monitored in real time by high-precision detection equipment installed on the printing press. The detection equipment collects the quality data of the printed products every 1 minute and compares it with the preset threshold. When it is detected that the color difference of a batch of packaging boxes reaches ΔE = 2.5, the early warning mechanism is triggered. After the early warning mechanism is triggered, an instruction is immediately sent to suspend the currently running printing press. However, since this batch of printing tasks is very urgent and the customer demand is urgent, the system automatically switches to the pre-set backup printing mode. The backup mode uses optimized and verified preset printing parameters, such as adjusting the ink formula and increasing printing pressure. At the same time, for packaging boxes that have been printed but do not meet the quality standards, they are automatically marked by a laser marking device. On the conveyor belt, mechanical push rods are used to push the marked unqualified products to a separate channel to isolate them from qualified products, making it easier for them to be reworked or processed separately. This process not only prevents the production of more unqualified products in a timely manner, but also ensures the continuation of urgent tasks. At the same time, unqualified products are effectively managed and processed, minimizing losses and waste and meeting customer needs and quality requirements.
[0045] Specifically, in some possible embodiments, the method includes: Collect various data from the print quality database, including image data, test results, printing machine parameters, environmental data, etc., and integrate and standardize them; Eliminate noise, missing values and outliers in the data to ensure the accuracy and completeness of the data; Extracting features from the data, including but not limited to color distribution features of statistical images and change trend features of printing press parameters; Select a preset big data analysis model and apply it to data mining tasks to obtain mining results; Analyze and interpret the mining results to derive potential quality problem patterns, key factors related to quality, and possible improvement directions; The mining results are visualized.
[0046] First, various types of data are widely collected from the print quality database established by the enterprise. This includes high-definition image data of different batches of prints in the past few months, each test result (such as color deviation value, clarity score, overprint accuracy, etc.), various parameters of the printing press during the production process (such as pressure, speed, temperature, etc.), and environmental data at the time (such as workshop temperature, humidity, lighting conditions, etc.). The collected data are integrated and standardized. For example, the test results of different formats and units are uniformly converted into standard formats and units for subsequent analysis. Data cleaning technology is used to eliminate noise, missing values and outliers in the data. Some test data that obviously deviate from the normal range or records with missing key parameters are deleted or supplemented. Then, valuable features are extracted from the integrated data. For example, the frequency and distribution of different colors in the image are counted to evaluate the color balance; the trend of the speed and pressure parameters of the printing press over time is analyzed to find the laws that may affect the quality. Then, a suitable big data analysis model, such as decision tree, cluster analysis, etc., is selected and applied to the data mining task. Through model operation, the mining results are obtained. Professional data analysts conduct in-depth analysis and interpretation of the mining results. They found that when the humidity in the workshop is high, the color of the printed product is prone to deviation; when the printing press speed is too fast, the accuracy of overprinting will decrease. Based on these findings, potential quality problem patterns and key factors related to quality are identified. Finally, these mining results are visualized through intuitive charts (such as bar charts, line charts, heat maps, etc.). On the large screen of the company's quality control center, the impact of different environmental conditions and printing press parameters on printing quality is clearly displayed, providing production personnel and managers with a clear basis for decision-making. For example, through visual display, the production team found that the quality of printed products produced in a certain month was generally poor. After further analysis of the data, it was found that it was due to the large temperature fluctuations in the workshop that month. Based on this, measures to strengthen temperature control were taken, which significantly improved the quality of subsequent printed products. In this way, the data in the print quality database is effectively utilized, providing strong support for the continuous improvement of printing quality.
[0047] Specifically, in some possible embodiments, the method includes: Based on the analysis results of the printed product quality database and the identification of key factors, the adjustment plan of the printing press parameters is automatically generated using machine learning algorithms; Pushing the adjustment plan to the terminal of the printing press operator in real time through a human-computer interaction interface or a mobile terminal application, while providing a detailed description of the plan and the expected effect; Obtain the corresponding parameter adjustments and operations performed by the operator on the printing press control system; The system automatically collects actual printing quality data and equipment operating status data after the operator implements the adjustment plan; Based on the actual printing quality data and equipment operation status data, combined with the new detection results of the collected images, the original adjustment plan is optimized and improved to improve the accuracy and effectiveness of the plan; With the continuous advancement of the production process and the continuous accumulation of data, the quality feedback mechanism is continuously updated and improved to adapt to different production conditions and quality requirements.
[0048] First, the print quality database was deeply analyzed to identify key factors affecting print quality, such as the drying speed of ink and the water absorption of paper. Machine learning algorithms, such as the random forest algorithm, were used to automatically generate adjustment plans for printing press parameters based on these key factors and historical data. For example, for a specific type of paper and ink combination, it is recommended to adjust the printing pressure to X Newtons and the printing speed to Y meters per minute. The adjustment plan was pushed to the printing press operator in real time through the human-machine interface in the printing workshop and the operator's mobile terminal application. The push content not only includes the specific parameters of the adjustment plan, but also provides detailed instructions, such as "adjusting the printing pressure can improve the ink penetration effect and is expected to increase the pattern clarity by 10%." After receiving the push, the operator adjusted the parameters and operated according to the plan on the printing press control system. The system automatically collects the actual printing quality data after the operator implements the adjustment plan, such as the measurement value of color accuracy, the resolution of the pattern, etc., as well as the operating status data of the equipment, such as temperature, humidity, energy consumption, etc. Combined with the newly collected image detection results, it was found that although the pattern clarity has improved, the color accuracy has not yet reached the ideal level. Based on these new data, the original adjustment scheme is optimized and improved, such as further fine-tuning the printing pressure or changing the ink type. As production continues and data continues to accumulate, the quality feedback mechanism is also continuously updated and improved. For example, when a new paper material or ink brand is introduced, the system can automatically adjust the weights and algorithm models of key factors to generate an adjustment scheme that is more suitable for the new production conditions. Through this implementation process, the quality of printed products has been steadily improved, production efficiency has also been improved, and it can better adapt to the ever-changing production needs and quality requirements.
[0049] Another embodiment of the present application provides a real-time monitoring system for a printing press based on image processing, wherein referring to Figure 2 , a real-time monitoring system for a printing press based on image processing, comprising: The static image and dynamic video stream acquisition module 100 is used to acquire static images and dynamic video streams of printed products at different angles and light conditions for printed products at various printing stages on a high-speed printing production line through multiple types of image acquisition devices provided in the printing press; The static image and dynamic video stream screening module 200 is used to screen the static image and the dynamic video stream to obtain the captured image; The collected image preprocessing module 300: for complex color prints, uses specific image enhancement technology and color space conversion technology to respectively enhance the clarity and contrast of the collected image, and converts the collected image into a corresponding color space; The detection frame division and adjustment module 400 uses a target detection algorithm in deep learning to preliminarily identify and locate the elements on the printed matter, and then adaptively divides the detection frame according to the characteristics of the elements, and dynamically adjusts the position and size of the detection frame in real time; Relevant edge information acquisition module 500: for the pre-processed collected image, using multiple edge detection algorithms, correspondingly obtaining multiple sets of edge detection results, and obtaining relevant edge information through a fusion strategy; Printing quality evaluation module 600: uses the optimization algorithm in machine learning to calculate the optimal grayscale threshold range, and evaluates the printing quality of the printed product under different ambient lights based on the optimal grayscale threshold range to obtain an evaluation result.
[0050] The present embodiment provides a real-time monitoring system for a printing press based on image processing. Due to the functions of each module itself and the logical connections between each other, it can implement the various steps of the aforementioned embodiment, and thus can achieve the same technical effects as the aforementioned embodiment. The principle analysis can refer to the relevant description of the steps of the aforementioned real-time monitoring method for a printing press based on image processing, which will not be repeated here.
[0051] An embodiment of the present application further provides a real-time monitoring device for a printing press based on image processing, comprising a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and execute the above-mentioned real-time monitoring method for a printing press based on image processing.
[0052] An embodiment of the present application further provides a storage medium, in which a computer program capable of being loaded by a processor and executing the above-mentioned real-time monitoring method of a printing press based on image processing is stored.
[0053] The storage medium provided in this embodiment can achieve the same technical effects as the aforementioned embodiments because the computer program therein implements the various steps of the aforementioned embodiments after being loaded and run on the processor. The principle analysis can be found in the relevant description of the aforementioned method steps, which will not be repeated here.
[0054] The storage medium includes, for example, various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.
[0055] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
[0056] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.
[0057] In addition, the term "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, "plurality" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined, and is only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features.
[0058] Thus, any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present invention belong.
[0059] The embodiments of this specific implementation method are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Therefore, all equivalent changes made based on the structure, shape, and principle of the present application should be included in the protection scope of the present application.
Claims
1. A real-time monitoring method for a printing press based on image processing, It is characterized in that include: On a high-speed printing production line, the printer uses multiple types of image acquisition devices to obtain static images and dynamic video streams of printed products at different angles and lighting conditions at each printing stage. Based on the static image and the dynamic video stream, screening is performed to obtain a captured image; For complex color prints, specific image enhancement technology and color space conversion technology are used to respectively enhance the clarity and contrast of the captured image, and convert the captured image into a corresponding color space; Using the target detection algorithm in deep learning, the elements on the printed matter are preliminarily identified and located, and then the detection frame is adaptively divided according to the characteristics of the elements, and the position and size of the detection frame are dynamically adjusted in real time; For the preprocessed collected image, multiple edge detection algorithms are used to obtain multiple sets of edge detection results accordingly, and relevant edge information is obtained through a fusion strategy; The optimization algorithm in machine learning is used to calculate the optimal grayscale threshold range, and the printing quality of the printed product under different ambient lights is evaluated based on the optimal grayscale threshold range to obtain an evaluation result.
2. According to the method for real-time monitoring of a printing press based on image processing as described in claim 1, It is characterized in that The method also covers: Build a real-time communication link between the detection system and the printing press control system; When it is detected that the printing quality of the printed product exceeds a preset threshold, in a large-scale printing production scenario, an early warning signal is sent and relevant parameters of the printing press are automatically adjusted.
3. According to the method for real-time monitoring of a printing press based on image processing as described in claim 2, It is characterized in that The method further includes: Create a print quality database to store each captured image, test results, and relevant parameters for printing press adjustment; Conduct regular analysis on the data in the printed product quality database, extract trends and rules of quality changes, and obtain analysis results; Based on the analysis results, identifying key factors affecting the quality of printed products; When the captured image is acquired, a printing adjustment plan is automatically generated based on the key factors.
4. According to the method for real-time monitoring of a printing press based on image processing as described in claim 1, It is characterized in that Before applying specific image enhancement technology and color space conversion technology to complex color prints, the following steps are included: Acquire images of prints of varying types, colors, and pattern complexity; Annotating the printed matter image to obtain annotation data, wherein the annotation data includes information such as the type, color characteristics, and pattern complexity of the printed matter; Build a multi-layer convolutional neural network model, set the convolution layer, pooling layer and fully connected layer, and determine the network structure and parameters; Using the labeled data to train the multi-layer convolutional neural network model, and continuously adjusting the weights and biases of the multi-layer convolutional neural network model through a back propagation algorithm to optimize model performance; Use the validation set to evaluate the trained multi-layer convolutional neural network model, calculate the accuracy, recall and other indicators, and adjust the structure or parameters if the requirements are not met until the evaluation result of the multi-layer convolutional neural network model is good; The collected printed images are input into the trained multi-layer convolutional neural network model to obtain the image classification results, and the printed images are classified according to the type, color, pattern complexity and other attributes of the printed materials.
5. According to the method for real-time monitoring of a printing press based on image processing as described in claim 3, It is characterized in that When it is detected that the printing quality of the printed product exceeds a preset threshold, the method comprises the following steps: According to the quality standards and production requirements of printed materials, pre-set the threshold of printing quality, including the upper and lower limits of various key indicators; Continuously monitor the printing quality of printed products in real time, compare the detected quality data with a preset threshold, and trigger an early warning mechanism when the quality data exceeds the preset threshold; When the warning is triggered, a command is sent to suspend the operation of the printing press; If the printing task is urgent, the system automatically switches to the pre-set backup printing mode, which uses the preset printing parameters; The products that have been printed but do not meet the quality standards are automatically marked, and the unqualified products are separated from the qualified products through mechanical devices or transmission systems for subsequent separate processing.
6. According to the method for real-time monitoring of a printing press based on image processing as described in claim 3, It is characterized in that The approach covers: Collect various data from the print quality database, including image data, test results, printing machine parameters, environmental data, etc., and integrate and standardize them; Eliminate noise, missing values and outliers in the data to ensure the accuracy and completeness of the data; Extracting features from the data, including but not limited to color distribution features of statistical images and change trend features of printing press parameters; Select a preset big data analysis model and apply it to data mining tasks to obtain mining results; Analyze and interpret the mining results to derive potential quality problem patterns, key factors related to quality, and possible improvement directions; The mining results are visualized.
7. According to the method for real-time monitoring of a printing press based on image processing as claimed in claim 3, It is characterized in that The method includes: Based on the analysis results of the printed product quality database and the identification of key factors, the adjustment plan of the printing press parameters is automatically generated using machine learning algorithms; Pushing the adjustment plan to the terminal of the printing press operator in real time through a human-computer interaction interface or a mobile terminal application, while providing a detailed description of the plan and the expected effect; Obtain the corresponding parameter adjustments and operations performed by the operator on the printing press control system; The system automatically collects actual printing quality data and equipment operating status data after the operator implements the adjustment plan; Based on the actual printing quality data and equipment operation status data, combined with the new detection results of the collected images, the original adjustment plan is optimized and improved to improve the accuracy and effectiveness of the plan; With the continuous advancement of the production process and the continuous accumulation of data, the quality feedback mechanism is continuously updated and improved to adapt to different production conditions and quality requirements.
8. A real-time monitoring system for printing presses based on image processing. It is characterized in that include: Static image and dynamic video stream acquisition module: used on a high-speed printing production line, through the multi-type image acquisition devices set up by the printing press, to obtain static images and dynamic video streams of the printed products at different angles and light conditions for the printed products at various printing stages; Static image and dynamic video stream screening module: based on the static image and the dynamic video stream, screening is performed to obtain the captured image; Captured image preprocessing module: for complex color prints, specific image enhancement technology and color space conversion technology are used to respectively improve the clarity and contrast of the captured image, and convert the captured image to the corresponding color space; Detection frame division and adjustment module: using the target detection algorithm in deep learning to preliminarily identify and locate the elements on the printed matter, then adaptively dividing the detection frame according to the characteristics of the elements, and dynamically adjusting the position and size of the detection frame in real time; Related edge information acquisition module: for the pre-processed collected image, multiple edge detection algorithms are used to obtain multiple sets of edge detection results accordingly, and related edge information is obtained through a fusion strategy; Printing quality assessment module: uses the optimization algorithm in machine learning to calculate the optimal grayscale threshold range, and evaluates the printing quality of the printed product under different ambient lights based on the optimal grayscale threshold range to obtain an evaluation result.
9. A real-time monitoring device for printing presses based on image processing, It is characterized in that include: A memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and execute any one of the image processing-based real-time monitoring methods for a printing press according to claims 1-7.
10. A storage medium, It is characterized in that The computer program is stored which can be loaded by a processor and execute any one of the image processing-based real-time monitoring methods for printing presses according to claims 1-7.
Citation Information
Cited By
Machine vision method and system for self-adaptive ink supply quantity optimization in gravure printing of decorative paper
CN119335876A
Control method and system for non-woven fabric production line
CN119511989A
Pressure feedback analysis method and system applied to visual printing machine
CN119559188A
Analysis method and system for printing performance of instant billing
CN119624257A
Panel printing monitoring method and device based on Internet of Things and medium
CN119644947A
