Method and system for monitoring running state of printing equipment

Abstract

The invention relates to the field of image data processing, in particular to a printing equipment running state monitoring method and system, and the method comprises the steps: carrying out the frame extraction of an obtained real-time image of a printed matter, obtaining a frame image, layering the image according to the image information entropy, and obtaining a real-time edge image layer, a real-time color filling image layer and a real-time background image layer; setting a standard image, obtaining an edge image layer of the standard image and taking the edge image layer as a standard edge image layer, and calculating the similarity between the real-time edge image layer and the standard edge image layer; calculating the stability of the real-time color filling layer and the stability of the real-time background layer; in the operation process of the printing equipment, the similarity and the stability are input into a preset monitoring model, monitoring results are output, and the detection results comprise flawless, color flaws, burr flaws and color and burr flaws. The method has the effect of improving the quality evaluation comprehensiveness of the printed matter.

Description

Technical Field

[0001] This application relates to the field of image data processing, and in particular, to a method and system for monitoring the operating state of a printing device. Background Art

[0002] The process of a printing press printing labels is a technology involving multiple steps, including: designing and making label patterns; making printing plates or printing templates; selecting appropriate label materials, such as paper, plastic film, metal foil, etc.; pouring ink into the ink cartridge of the printing press, or automatically inking during the printing process; starting the printing press, passing the label material through the printing press, and transferring the ink to the material by the printing plate or print head.

[0003] Printed products may have various quality problems, such as uneven coloring or color difference, burrs in printing, etc. These reasons may be caused by uneven ink supply, uneven printing pressure, or improper position of the rubber roller. Therefore, quality monitoring of printed products is required to reflect the operating state of the printing device. In the prior art, defects of printed products can be identified through image processing. For example, the Chinese patent application "Method and System for Real-time Monitoring of a Printing Press Based on Image Processing" with the publication number CN117830315A discloses: processing a printed product image and obtaining a grayscale image after partitioning with detection frames, performing edge detection to obtain an edge image, and calculating the gradient values of pixel points within the connected regions in the edge image; calculating the smudging degree of any detection frame when any grayscale value is used as a threshold; clustering the smudging degree sequence and the center coordinates of the detection frames to obtain multiple clustering categories and performing binary classification to obtain smudging area categories and normal printing area categories; calculating the threshold segmentation effect; taking the threshold corresponding to the maximum value of the threshold segmentation effect as the optimal segmentation threshold, and performing image segmentation on the grayscale image according to the optimal segmentation threshold and inputting it into a preset monitoring model to generate a defect monitoring result.

[0004] In the above solution, only the smudging situation is identified. The quality of printed products is affected by multiple factors, including color, edge sharpness, etc. Single-dimensional detection may not be able to comprehensively capture these factors, resulting in incomplete quality assessment. Summary of the Invention

[0005] To solve the problem that single-dimensional defect detection has limitations in the quality assessment of printed products, this application provides a method and system for monitoring the operating state of a printing device.

[0006] In a first aspect, this application provides a method for monitoring the operating state of a printing device, adopting the following technical solution: A method for monitoring the operating state of a printing device, comprising the steps of: extracting frames from the real-time image of the printed matter to obtain frame images, performing grayscale processing on the frame images to obtain grayscale images, dividing the grayscale images into multiple regions using grid lines, setting a window for each region, and calculating the image information entropy within the window; clustering the image information entropy to obtain multiple clustering clusters, such that the image information entropy within the same clustering cluster corresponds to the images within the window on the same layer of the image, to obtain a real-time edge layer, a real-time coloring layer, and a real-time background layer; setting a standard image, and using the same method as for obtaining the real-time edge layer to obtain the edge layer of the standard image and using it as the standard edge layer, and calculating the similarity between the real-time edge layer and the standard edge layer; calculating the stability of the real-time coloring layer and the stability of the real-time background layer; during the operation of the printing device, inputting the similarity and the stability into a preset monitoring model, and outputting a monitoring result, the detection result including: no defect, color defect, burr defect, color and burr defect.

[0007] The beneficial effects are as follows: clustering the image information entropy, and the images corresponding to the image information entropy within the same clustering cluster within the window can be considered to be on the same layer of the image. The edge layer, the coloring layer, and the background layer can be distinguished. Real-time edge layer: Usually, the edge layer corresponds to the region with a higher information entropy because the edge region contains more details and variations. Real-time coloring layer: The coloring layer corresponds to the region with a medium information entropy, and these regions usually contain relatively consistent colors or textures. Real-time background layer: The background layer may correspond to the region with a lower information entropy because the background region is usually more uniform and has little variation.

[0008] For different layers, calculate the similarity or stability according to the pattern characteristics of the layer to identify the corresponding defects, and achieve the purpose of monitoring the operating state of the printing device by identifying the defects of the printed matter. Identify multi-dimensional quality defects such as color defects and burr defects in printing through the monitoring model, so as to facilitate comprehensive quality evaluation of the printed matter.

[0009] Optionally, to calculate the similarity between the real-time edge layer and the standard edge layer, the calculation method is as follows: set a direction, scan the real-time edge layer to obtain the run length of the real-time edge layer, the run length refers to the length of consecutive pixels with the same grayscale level in the image, the run length of the real-time edge layer includes long run lengths and short run lengths, calculate the proportion of the short run lengths in all the run lengths of the real-time edge layer and use it as the first ratio; set a standard image, and use the same method as for obtaining the real-time edge layer to obtain the edge layer of the standard image and use it as the standard edge layer; similarly, obtain the run length of the standard edge layer and the proportion of the short run lengths and use it as the second ratio; calculate the similarity between the real-time edge layer and the standard edge layer according to the first ratio and the second ratio, and the similarity calculation formula is: , where, represents the similarity between the real-time edge layer and the standard edge layer, represents the first ratio, represents the second ratio, represents the exponential function with as the base.

[0010] The beneficial effect is that long runs correspond to large regions in the image, while short runs correspond to small or detailed regions. When there are burrs on the edge, multiple short runs will appear. The similarity between layers is quantified through the similarity calculation formula. When and the difference is small, tends to 0, the value of tends to 1. Therefore, being close to 1 indicates a high similarity between the real-time edge layer and the standard edge layer. Conversely, it indicates a low similarity between the real-time edge layer and the standard edge layer.

[0011] Optionally, to calculate the similarity between the real-time edge layer and the standard edge layer, the calculation method is as follows: construct the gray-level run-length matrix of the real-time edge layer; in the gray-level run-length matrix, the number of rows represents the gray levels in the image, and the number of columns represents the run length; similarly, construct the gray-level run-length matrix of the standard edge layer; calculate the energy value of the gray-level run-length matrix in the real-time edge layer, and the expression of the energy value is: ; in the formula, represents the energy value, represents the run length, represents the th row and the th column element value, represents the total number of represents the total number of ; similarly, calculate the corresponding energy value of the standard edge layer; the similarity calculation formula is: , where represents the similarity between the real-time edge layer and the standard edge layer, represents the energy value of the gray-level run-length matrix in the real-time edge layer, represents the corresponding energy value of the standard edge layer, represents the exponential function with as the base.

[0012] The beneficial effect is that when and the difference is small, the similarity Close to 1, which indicates a relatively high similarity between the real-time edge layer and the standard edge layer. Conversely, the similarity between the real-time edge layer and the standard edge layer is relatively low. Burrs or unclear edges occur during the printing process. Burrs may cause the run length in the image to become shorter, thereby reducing the energy value. The similarity between the layers is quantified by the above calculation formula.

[0013] Optionally, calculate the similarity between the real-time edge layer and the standard edge layer. The calculation formula for the similarity is: , where represents the similarity between the real-time edge layer and the standard edge layer, is the number of windows in the standard edge layer, is the number of windows in the real-time edge layer, is the information entropy of window in the real-time edge layer, is the information entropy of window in the real-time edge layer, represents the exponential function with as the base.

[0014] The beneficial effect is that when the number of windows and the information entropy distribution of the two layers are very close, the similarity is close to 1, indicating that the two layers are very similar. If the difference in the number of windows is large or the difference in the average value of the information entropy is large, the similarity will decrease, indicating that the similarity between the two layers is relatively low.

[0015] Optionally, calculating the stability of the real-time filling layer and the stability of the real-time background layer includes the steps: For any window, calculate the multi-channel color difference between two pixel points corresponding to the position. The calculation formula for the multi-channel color difference of pixel points is: , where represents the multi-channel color difference between the pixel point on the real-time filling layer and the pixel point on the standard filling layer, represents the red color value of the pixel point , represents the red color value of the pixel point , represents the green color value of the pixel point , represents the green color value of the pixel point , represents the blue color value of the pixel point , represents the blue color value of the pixel point ​​​​​The blue color value; the mean of the multi-channel color differences of all pixel points is used as the stability of the real-time coloring layer; in the same way as the method for calculating the stability of the real-time coloring layer, calculate the stability of the real-time background layer.

[0016] The beneficial effect is that the lower the stability, the smaller the color difference between the real-time layer and the standard layer, that is, the closer the color distribution of the real-time layer is to the standard layer, and vice versa, indicating a larger color difference. By calculating the stability, the color quality of the printed matter is quantified to ensure that the filled color and the background color are consistent with the standard color, thus ensuring the overall quality of the printed matter.

[0017] Optionally, calculating the stability of the real-time coloring layer and the stability of the real-time background layer includes the steps of: separating the color image of the real-time coloring layer into a real-time red sub-layer, a real-time blue sub-layer, and a real-time green sub-layer according to color channels; similarly, obtaining the standard red sub-layer, the standard blue sub-layer, and the standard green sub-layer of the standard coloring layer; for any window, calculating the mean of the single-channel color differences of the pixel points on the real-time red sub-layer and the standard red sub-layer to obtain the red channel color difference; similarly obtaining the blue channel color difference and the green channel color difference; setting the first weight coefficient, the second weight coefficient, and the third weight coefficient, and taking the mean of the sum of the product of the first weight coefficient and the red channel color difference, the product of the second weight coefficient and the green channel color difference, and the product of the third weight coefficient and the blue channel color difference as the stability of the real-time coloring layer; in the same way as the method for calculating the stability of the real-time coloring layer, calculate the stability of the real-time background layer.

[0018] The beneficial effect is that through this weighted calculation, the stability of the layer can be more accurately evaluated, especially in the detection of printed matter where color quality has a greater visual impact. The weighting method can highlight the more important color channels, making the calculation result of the stability more in line with the actually observed effect, thereby improving the accuracy and reliability of print quality control.

[0019] Optionally, the setting method of the monitoring model is: setting a data set; the setting method of the data set is: in the same way as calculating the similarity between the real-time edge layer and the standard edge layer, calculate the similarity between the historical edge layer of the historical frame image and the standard edge layer; in the same way as calculating the stability of the real-time coloring layer and the stability of the real-time background layer, calculate the stability of the historical coloring layer and the stability of the historical background layer of the historical frame image; setting a multi-dimensional data vector regarding the window number, similarity, stability of the real-time coloring layer, and stability of the real-time background layer; tagging the multi-dimensional data vector, and the tags are flawless, color flaw, burr flaw, color and burr flaw; training a neural network model according to the data set, and stopping training to obtain the monitoring model when the preset training stop condition is reached.

[0020] Second aspect, the present application provides a monitoring system for the operating state of a printing device, adopting the following technical solution: A monitoring system for the operating state of a printing device, comprising: a processor and a memory, the memory storing computer program instructions, which when executed by the processor implement the monitoring method for the operating state of the printing device as described above.

[0021] The beneficial effect is that: the above-mentioned monitoring method for the operating state of the printing device is generated into a computer program and stored in the memory to be loaded and executed by the processor. Thus, a system is made according to the memory and the processor, which is convenient to use.

[0022] The present application has the following technical effects: 1. Stratify according to the image information entropy. For different layers, calculate the similarity or stability according to the pattern characteristics of the layer to identify the corresponding defects, and achieve the purpose of monitoring the operating state of the printing device by identifying the defects of the printed matter. Identify multi-dimensional quality defects such as color defects and burr defects in printing through the monitoring model, so as to facilitate the comprehensive quality evaluation of the printed matter.

[0023] 2. Cluster the image information entropy. The images within the corresponding window of the image information entropy within the same clustering cluster can be considered to be on the same layer of the image. The edge layer, the coloring layer, and the background layer can be distinguished. Real-time edge layer: Usually, the edge layer corresponds to the area with higher information entropy because the edge area contains more details and variations. Real-time coloring layer: The coloring layer corresponds to the area with medium information entropy, and these areas usually contain relatively consistent colors or textures. Real-time background layer: The background layer may correspond to the area with lower information entropy because the background area is usually more uniform and has little change. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] By referring to the drawings and reading the following detailed description, the above and other objects, features, and advantages of the exemplary embodiments of the present application will become easily understandable. In the drawings, several embodiments of the present application are shown in an exemplary rather than restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts.

[0025] Figure 1 It is a flowchart of a method for monitoring the operating state of a printing device according to an embodiment of the present application.

[0026] Figure 2 It is a schematic structural diagram showing the printing base paper and the printing pattern in a method for monitoring the operating state of a printing device according to an embodiment of the present application.

[0027] Figure 3 It is a flowchart of a method for calculating similarity in a method for monitoring the operating state of a printing device according to an embodiment of the present application.

[0028] Figure 4 It is a flowchart of another similarity calculation method in the method for monitoring the operating state of a printing device according to an embodiment of the present application.

[0029] Explanation of reference numerals: 1, printing base paper; 2, label pattern. Detailed implementation manners

[0030] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts shall fall within the protection scope of the present application.

[0031] It should be understood that when the claims, the description and the drawings of the present application use terms such as "first" and "second", they are only used to distinguish different objects, rather than to describe a specific order. The terms "including" and "comprising" used in the description and claims of the present application indicate the existence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the existence or addition of one or more other features, wholes, steps, operations, elements, components and / or their combinations.

[0032] An embodiment of the present application discloses a method for monitoring the operating state of a printing device. Referring to Figure 1 , it includes steps S1 - S5, specifically as follows: S1: Extract frames from the real-time image of the printed matter to obtain frame images, perform grayscale processing on the frame images to obtain grayscale images, use grid lines to divide the grayscale images into multiple regions, set a window for each region, and calculate the image information entropy within the window.

[0033] In one embodiment, the method for frame extraction is: real-time monitor the conveying speed of the label paper through a sensor, and then dynamically adjust the frame extraction frequency according to the speed change to ensure that each label paper can be clearly displayed on different frames. Specifically, referring to Figure 2 , a plurality of label patterns 2 (label papers) need to be continuously printed on each conveyed printing base paper 1. The setting of the frame extraction frequency needs to consider the conveying speed of the label paper to ensure that each label paper can be clearly displayed on different frames. The first label paper is on the first frame image, and the second label paper is on the second frame image. The distance between the first label paper and the second label paper ( Figure 2The ratio of d) in to the transmission speed is used as the period, and the reciprocal of the period is used as the frame extraction frequency. It is also possible to extract the label paper at intervals for printing detection. For example, one label paper is extracted for monitoring every 10 label papers. Just multiply the above distance by 10 and then calculate the frame extraction frequency. 10 is a hyperparameter and can be adjusted by the user according to the specific application scenario. Dynamically adjusting the frame extraction frequency according to the transmission speed of the label paper can reduce the problems of image overlap or omission caused by a fixed frame extraction frequency.

[0034] In other embodiments, the frame extraction frequency can also be adjusted by the user according to the specific usage scenario, which will not be elaborated here.

[0035] Draw grid lines on the grayscale image to divide the grayscale image into multiple regions. Exemplarily, the size of the grid lines can be 8×12, where 8 is the number of rows of the grid lines and 12 is the number of columns of the grid lines; it can also be 10×10. The size of the grid lines can be set by the user according to the specific application scenario.

[0036] The grid lines divide the grayscale image into multiple equal-sized rectangular regions. Use the length and width of the rectangular region as the length and width of the window to set the rectangular window. Calculate the histogram for the grayscale image within each window. The histogram shows the frequency distribution of different grayscale values. Calculate the image information entropy of each window: , where represents the image information entropy, is the th normalized probability of the gray level (that is, the frequency of the th gray level in the histogram divided by the total number of pixels), is the number of gray levels (usually 256).

[0037] S2: Cluster the image information entropy to obtain multiple clusters, so that the images corresponding to the image information entropy within the same cluster are on the same layer of the image, obtaining a real-time edge layer, a real-time filling layer, and a real-time background layer.

[0038] Exemplarily, the clustering algorithm can adopt the K-means clustering algorithm or DBSCAN (Density-Based Spatial Clustering of Applications with Noise, density-based spatial clustering algorithm). Both the K-means clustering algorithm and DBSCAN are existing technologies and will not be elaborated here.

[0039] After clustering, the image information entropy within the same clustering cluster corresponding to the window can be considered to be on the same layer of the image. According to the clustering results, the edge layer, the coloring layer, and the background layer can be distinguished. Real-time edge layer: Usually, the edge layer corresponds to the area with higher information entropy because the edge area contains more details and variations. Real-time coloring layer: The coloring layer corresponds to the area with medium information entropy, and these areas usually contain relatively consistent colors or textures. Real-time background layer: The background layer may correspond to the area with lower information entropy because the background area is usually relatively uniform and has little variation.

[0040] S3: Set a standard image. In the same way as the method for obtaining the real-time edge layer, obtain the edge layer of the standard image and use it as the standard edge layer, and calculate the similarity between the real-time edge layer and the standard edge layer.

[0041] In one embodiment, the calculation method of the similarity includes step S30 - step S32, which is as follows: S30: Set a direction, scan the real-time edge layer to obtain the run length of the real-time edge layer. The run length refers to the length of consecutive pixels with the same gray level in the image. The run length of the real-time edge layer includes long run lengths and short run lengths. Calculate the proportion of the short run lengths in all the run lengths of the real-time edge layer and use it as the first ratio.

[0042] Exemplary directions can be the horizontal direction or the vertical direction. Long run lengths correspond to large areas in the image, while short run lengths correspond to small or detailed areas. When there are burrs on the edge, multiple short run lengths will appear. A long run length is a sequence of consecutive pixels with the same gray level whose length is greater than or equal to a preset threshold. A sequence of consecutive pixels with the same gray level whose length is less than the preset threshold is a short run length. Exemplarily, the threshold is set to 3, and the setting of the threshold can be set by the user according to the actual application scenario.

[0043] S31: Set a standard image. In the same way as the method for obtaining the real-time edge layer, obtain the edge layer of the standard image and use it as the standard edge layer; similarly, obtain the run length of the standard edge layer and the proportion of short run lengths and use it as the second ratio.

[0044] S32: Calculate the similarity between the real-time edge layer and the standard edge layer according to the first ratio and the second ratio.

[0045] The similarity calculation formula is: , where represents the similarity between the real-time edge layer and the standard edge layer, represents the first ratio, represents the second ratio. represents the exponential function with as the base, represents the exponential function with The exponential function with a certain base.

[0046] Among them, when and the difference is small, tends to 0, the value of tends to 1. Therefore, being close to 1 indicates a high similarity between the real-time edge layer and the standard edge layer. On the contrary, it indicates a low similarity between the real-time edge layer and the standard edge layer.

[0047] In one embodiment, the method for calculating the similarity includes steps S33 - S35, specifically as follows: S33: Construct the gray-level run-length matrix of the real-time edge layer; in the gray-level run-length matrix, the number of rows represents the gray levels in the image, and the number of columns represents the run length; similarly, construct the gray-level run-length matrix of the standard edge layer.

[0048] The method for constructing the gray-level run-length matrix is: Scan each pixel of the real-time edge layer and record the continuous length of each gray level, that is, the run. The number of rows of the gray-level run-length matrix corresponds to the number of possible gray levels in the image (for example, an 8-bit gray image has 256 gray levels), and the number of columns corresponds to the possible run lengths (usually determined according to the image size and content). For each combination of gray level and run length, calculate the number of occurrences in the image and fill it into the corresponding matrix element.

[0049] S34: Calculate the energy value of the gray-level run-length matrix in the real-time edge layer.

[0050] The expression for the energy value is: ; in the formula, represents the energy value, represents the run length, represents the th element value in the th column, represents the total number of represents the total number of

[0051] The higher the energy value, the more complex the texture in the image. This is because a complex texture will have more runs of different lengths, resulting in an increase in the term in the energy value.

[0052] S35: Calculate the similarity based on the energy value.

[0053] The formula for calculating the similarity is: , where represents the similarity between the real-time edge layer and the standard edge layer, represents the energy value of the gray-level run-length matrix in the real-time edge layer, represents the corresponding energy value of the standard edge layer, represents the exponential function with as the base.

[0054] When and have a small difference, the similarity is close to 1, indicating a high similarity between the real-time edge layer and the standard edge layer. Conversely, the similarity between the real-time edge layer and the standard edge layer is low.

[0055] It should be noted that in this embodiment, when is less than , it means that the energy value of the real-time edge layer is low, which may be due to burrs or unclear edges during the printing process. Burrs may cause the run-lengths in the image to become shorter, thus reducing the energy value. But when is greater than , it means that the energy value of the real-time edge layer is high, which may be because two line segments that should not be connected are connected during the printing process, resulting in longer run-lengths and increased energy value. This situation can be separately labeled as ink bleeding in the subsequent step S5.

[0056] In one embodiment, the similarity can also be calculated based on window data and information entropy. The calculation formula for the similarity is: , where represents the similarity between the real-time edge layer and the standard edge layer, is the number of windows in the standard edge layer, is the number of windows in the real-time edge layer, is the information entropy of window in the real-time edge layer, is the information entropy of window in the real-time edge layer, represents the exponential function with as the base.

[0057] expresses the difference in the number of windows between the two layers. When and have a small difference, that is, the number of windows in the two layers is the same, tends to 1, indicating that the number of windows is more similar. expresses the difference in the average information entropy of the windows between the two layers. When the average information entropy of the two layers is close, this value is close to 1, indicating similar information entropy distributions; the greater the difference, the smaller this value, indicating a larger difference in information entropy distributions.

[0058] When the number of windows and the information entropy distribution of two layers are very close, the similarity is close to 1, indicating that the two layers are very similar. If there is a large difference in the number of windows or a large difference in the average value of information entropy, the similarity will decrease, indicating a low similarity between the two layers.

[0059] S4: Calculate the stability of the real-time coloring layer and the stability of the real-time background layer.

[0060] In one embodiment, calculating the stability of the real-time coloring layer and the stability of the real-time background layer includes the steps of: For any window, calculate the multi-channel color difference between two pixel points with corresponding positions; take the average value of the multi-channel color differences of all pixel points as the stability of the real-time coloring layer; The calculation formula for the multi-channel color difference of pixel points is: , where represents the pixel point on the real-time coloring layer and the pixel point on the standard coloring layer of the multi-channel color difference, represents the pixel point of the red color value, represents the pixel point of the red color value, represents the pixel point of the green color value, represents the pixel point of the green color value, represents the pixel point of the blue color value, represents the pixel point of the blue color value.

[0061] The lower the stability, the smaller the color difference between the real-time layer and the standard layer, that is, the closer the color distribution of the real-time layer is to the standard layer. Conversely, it means a larger color difference. By calculating the stability, the color quality of the printed matter is quantified to ensure that the coloring and background colors are consistent with the standard colors, thus ensuring the overall quality of the printed matter.

[0062] Similarly to the method of calculating the stability of the real-time coloring layer, calculate the stability of the real-time background layer.

[0063] In one embodiment, the calculation method for calculating the stability of the real-time coloring layer and the stability of the real-time background layer can also be: Separate the color image of the real-time coloring layer into a real-time red sub-layer, a real-time blue sub-layer, and a real-time green sub-layer according to the color channels; similarly, obtain the standard red sub-layer, the standard blue sub-layer, and the standard green sub-layer of the standard coloring layer.

[0064] For any window, calculate the mean value of the single-channel color differences of the pixels on the real-time red sub-layer and the standard red sub-layer to obtain the red-channel color difference; similarly, obtain the blue-channel color difference and the green-channel color difference.

[0065] Set the first weight coefficient, the second weight coefficient, and the third weight coefficient, and take the mean value of the sum of the product of the first weight coefficient and the red-channel color difference, the product of the second weight coefficient and the green-channel color difference, and the product of the third weight coefficient and the blue-channel color difference as the stability of the real-time filling layer.

[0066] Exemplarily, set the first weight coefficient (red channel) to 0.3, set the second weight coefficient (green channel) to 0.5; set the third weight coefficient (blue channel) to 0.2. The setting of the weights can be set by the user according to the actual application scenario. Through this weighted calculation, the stability of the layer can be evaluated more accurately, especially in the detection of printed matter where color quality has a greater visual impact. The weighting method can highlight the more important color channels, making the calculation result of the stability more in line with the actually observed effect, thereby improving the accuracy and reliability of print quality control.

[0067] In the same way as the method for calculating the stability of the real-time filling layer, calculate the stability of the real-time background layer.

[0068] S5: During the operation of the printing device, input the similarity and stability into a preset monitoring model, and output the monitoring result. The detection results include: no defect, color defect, burr defect, color and burr defect.

[0069] Set the data set; train the neural network model according to the data set, and stop training to obtain the monitoring model when the preset training stop condition is reached.

[0070] In one embodiment, the method for setting the data set is as follows: In the same way as calculating the similarity between the real-time edge layer and the standard edge layer, calculate the similarity between the historical edge layer of the historical frame image and the standard edge layer; in the same way as calculating the stability of the real-time filling layer and the stability of the real-time background layer, calculate the stability of the historical filling layer and the stability of the historical background layer of the historical frame image; set a multi-dimensional data vector regarding the window number, similarity, stability of the real-time filling layer, and stability of the real-time background layer; label the multi-dimensional data vector, and the labels are no defect, color defect, burr defect, color and burr defect.

[0071] In one embodiment, the neural network model adopts a BP (BackPropagation) neural network model, the loss function adopts a mean squared error loss function, and the training stop condition is that the value of the loss function is less than 0.01 or the number of training times reaches 1000 times. The training of the model is prior art and will not be elaborated here.

[0072] By combining historical data and the neural network model, effective quality control of printed matter can be achieved, so as to monitor the operating state of the printing equipment. The defects of the printed matter reflect the printing defects of the printing equipment, such as ink problems, printing roller problems, etc. By identifying the defects of the printed matter, the user can be guided to conduct targeted troubleshooting and maintenance.

[0073] The embodiment of the present application also discloses a monitoring system for the operating state of a printing device, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the monitoring method for the operating state of the printing device according to the present application is implemented.

[0074] The above system also includes other components well-known to those skilled in the art, such as a communication bus and a communication interface. Their settings and functions are known in the art, so they will not be elaborated here.

[0075] In the present application, the foregoing memory can be any tangible medium that contains or stores a program, and the program can be used by or in combination with an instruction execution system, apparatus, or device. For example, the computer-readable storage medium can be any suitable magnetic storage medium or magneto-optical storage medium, such as resistive random access memory (RRAM), dynamic random access memory (DRAM), static random access memory (SRAM), enhanced dynamic random access memory (EDRAM), high bandwidth memory (HBM), hybrid memory cube (HMC), etc., or any other medium that can be used to store the required information and can be accessed by an application program, module, or both. Any such computer storage medium can be part of the device or accessible or connectable to the device.

[0076] Although this specification has shown and described multiple embodiments of the present application, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will think of many changes, alterations, and alternative ways without departing from the spirit and concept of the present application. It should be understood that various alternatives to the embodiments of the present application described herein may be employed in practicing the present application.

[0077] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereby. Therefore, all equivalent changes made according to the structure, shape, and principle of the present application shall be covered within the protection scope of the present application.

Claims

1. A method for monitoring the operating status of a printing device, characterized in that: Includes steps: Extract frames from the acquired real-time image of the printed matter to obtain a frame image, perform grayscale processing on the frame image to obtain a grayscale image, divide the grayscale image into multiple regions using grid lines, set a window in each region, and calculate the image information entropy in the window; Clustering the image information entropy to obtain multiple clusters, so that the image information entropy in the same cluster corresponds to the image in the window on the same layer of image, to obtain a real-time edge layer, a real-time coloring layer and a real-time background layer; Setting a standard image, in the same way as obtaining a real-time edge layer, obtaining an edge layer of the standard image and using it as the standard edge layer, and calculating the similarity between the real-time edge layer and the standard edge layer; Calculate the stability of the real-time fill layer and the stability of the real-time background layer; During the operation of the printing equipment, the similarity and stability are input into the preset monitoring model, and the monitoring results are output. The detection results include: no defects, color defects, burr defects, color and burr defects.

2. The method for monitoring the operating status of a printing device according to claim 1, characterized in that: Calculate the similarity between the real-time edge layer and the standard edge layer. The calculation method is: Set the direction, scan the real-time edge layer to obtain the run length of the real-time edge layer. The run length refers to the length of consecutive pixels with the same gray level in the image. The run length of the real-time edge layer includes long run length and short run length. Calculate the proportion of short run length in the run length of all real-time edge layers and use it as the first ratio; Setting a standard image is similar to the method of obtaining a real-time edge layer, and the edge layer of the standard image is obtained and used as the standard edge layer; similarly, the run length and short run length ratio of the standard edge layer are obtained and used as the second ratio; The similarity between the real-time edge layer and the standard edge layer is calculated according to the first ratio and the second ratio. The similarity calculation formula is: ,in, Indicates the similarity between the real-time edge layer and the standard edge layer. represents the first ratio, represents the second ratio, Indicates An exponential function with base .

3. The method for monitoring the operating status of a printing device according to claim 1, characterized in that: include: Calculate the similarity between the real-time edge layer and the standard edge layer. The calculation method is: Construct the grayscale run-length matrix of the real-time edge layer; In the grayscale run-length matrix, the number of rows represents the grayscale level in the image, and the number of columns represents the run-length. Similarly, the grayscale run-length matrix of the standard edge layer is constructed. Calculate the energy value of the grayscale run-length matrix in the real-time edge layer. The expression of the energy value is: ; In the formula, Indicates the energy value, represents the run length, Indicates Line The element value of the column, express The total number of express Similarly, calculate the energy value corresponding to the standard edge layer; The similarity calculation formula is: ,in, Indicates the similarity between the real-time edge layer and the standard edge layer. Represents the energy value of the grayscale run-length matrix in the real-time edge layer, Represents the energy value corresponding to the standard edge layer, Indicates An exponential function with base .

4. The method for monitoring the operating status of a printing device according to claim 1, characterized in that: Calculate the similarity between the real-time edge layer and the standard edge layer. The similarity calculation formula is: ,in, Indicates the similarity between the real-time edge layer and the standard edge layer. is the number of windows in the standard edge layer, is the number of windows of the real-time edge layer, For the real-time edge layer window The information entropy of For the real-time edge layer window The information entropy of Indicates An exponential function with base .

5. The method for monitoring the operating status of a printing device according to claim 1, characterized in that: Calculating the stability of the real-time fill layer and the stability of the real-time background layer includes the following steps: For any window, the multi-channel color difference between two pixels at the corresponding position is calculated. The calculation formula for the multi-channel color difference of the pixel point is: ,in, Represents the pixels on the live fill layer Pixels on the standard fill layer Multi-channel chromatic aberration, Represents pixel The red color value, Represents pixel The red color value, Represents pixel The green color value, Represents pixel The green color value, Represents pixel The blue color value, Represents pixel The blue color value of The stability of the real-time color filling layer is achieved by taking the average of the multi-channel color differences of all pixels; The stability of the real-time background layer is calculated in the same way as the stability of the real-time fill layer.

6. The method for monitoring the operating status of a printing device according to claim 1, characterized in that: Calculating the stability of the real-time fill layer and the stability of the real-time background layer includes the following steps: The color image of the real-time color filling layer is divided into a real-time red sublayer, a real-time blue sublayer and a real-time green sublayer according to the color channel; similarly, a standard red sublayer, a standard blue sublayer and a standard green sublayer of the standard color filling layer are obtained; For any window, the average value of the single-channel color difference between the real-time red sublayer and the standard red sublayer is calculated to obtain the red channel color difference; similarly, the blue channel color difference and the green channel color difference are obtained; The first weight coefficient, the second weight coefficient and the third weight coefficient are set, and the average of the product of the first weight coefficient and the color difference of the red channel, the product of the second weight coefficient and the color difference of the green channel, and the product of the third weight coefficient and the color difference of the blue channel is used as the stability of the real-time color filling layer; The stability of the real-time background layer is calculated in the same way as the stability of the real-time fill layer.

7. The method for monitoring the operating status of a printing device according to claim 1, characterized in that: The monitoring model is set up as follows: Set the data set; the method of setting the data set is: similar to calculating the similarity between the real-time edge layer and the standard edge layer, calculate the similarity between the historical edge layer and the standard edge layer of the historical frame image; similar to calculating the stability of the real-time coloring layer and the stability of the real-time background layer, calculate the stability of the historical coloring layer and the stability of the historical background coating of the historical frame image; Set multidimensional data vectors about window number, similarity, stability of real-time fill layer and stability of real-time background layer; Label the multidimensional data vectors, the labels being flawless, color flaws, burr flaws, and color and burr flaws; The neural network model is trained according to the data set, and when the preset training stop condition is reached, the training is stopped to obtain the monitoring model.

8. A printing equipment operation status monitoring system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the method for monitoring the operating status of a printing device according to any one of claims 1 to 7 is implemented.

Images