Adaptive control method and system for high-speed rotary printing machine based on dynamic monitoring

Abstract

The application provides a high-speed rotary printing machine adaptive control method and system based on dynamic monitoring, and relates to the technical field of industrial control. The method comprises the following steps: acquiring a web image in a high-speed printing process; identifying a connected fusion area formed between a reference line feature and an alignment mark due to dynamic diffusion of ink; reconstructing an anti-diffusion feature of the alignment mark to obtain an independent mark image; calculating a register offset between color plates; and generating and issuing a control instruction to a nozzle execution module to adaptively adjust the printing machine. The application solves the technical problem that, in the prior art, due to dynamic diffusion of ink in a high-speed printing environment, the shape of the alignment mark is distorted, the register offset is caused, and the printing quality is affected. Through real-time monitoring and adaptive adjustment, the adaptive control is always based on real errors for accurate correction, and the printing precision and quality in the high-speed printing process are significantly improved.

Description

Technical Field

[0001] This application relates to the field of industrial control technology, specifically to an adaptive control method and system for high-speed rotary printing presses based on dynamic monitoring. Background Technology

[0002] In high-speed rotary printing presses, registration accuracy is one of the core indicators determining print quality, typically relying on visual inspection of the alignment marks of each color plate to achieve closed-loop control. Existing visual inspection solutions generally calculate the offset between each color plate by extracting the alignment marks and then generating control commands to drive the actuators for correction. In actual high-speed printing conditions, especially for printing materials with high ink absorption such as uncoated paper, the ink will undergo uncontrollable dynamic diffusion in the paper fibers before it is fully cured. This causes morphological distortion of the edges of the registration marks, and even pixel-level connectivity and fusion with adjacent reference line features along the printing paper feed direction. The resulting visual artifacts are not caused by mechanical displacement. Existing conventional image processing methods are difficult to effectively separate the false pixels caused by diffusion from the real physical boundaries of the marks. If the distorted mark image is directly used as the detection benchmark, the calculated overprinting offset will inevitably be mixed with the error component introduced by ink diffusion. This will cause the detection signal to fail to reflect the mechanical overprinting deviation between the rollers. Not only will it fail to achieve accurate correction, but it will also cause the overprinting error to diverge, which will seriously restrict the production quality and operational stability of high-speed rotary printing equipment under complex conditions.

[0003] In summary, existing technologies suffer from the technical problem that the dynamic diffusion of ink in high-speed printing environments causes distortion of the alignment mark shape, leading to misregistration and further affecting printing quality. Summary of the Invention

[0004] The purpose of this application is to provide an adaptive control method and system for high-speed rotary printing presses based on dynamic monitoring, in order to solve the technical problem in the prior art where the dynamic diffusion of ink in a high-speed printing environment causes distortion of the alignment mark shape, resulting in misregistration and further affecting the printing quality.

[0005] To achieve the above objectives, this application provides an adaptive control method and system for high-speed rotary printing presses based on dynamic monitoring.

[0006] In a first aspect, this application provides an adaptive control method for a high-speed rotary printing press based on dynamic monitoring. This method is implemented through an adaptive control system for a high-speed rotary printing press based on dynamic monitoring. The method includes: acquiring a web image during the high-speed printing process, the web image containing reference line features and alignment marks; identifying a connected and fused region between the reference line features and the alignment marks due to dynamic ink diffusion based on the web image; reconstructing anti-diffusion features of the alignment marks based on the connected and fused region to obtain an independent mark image; calculating the registration offset between each color plate based on the independent mark image; and generating and issuing control commands to the printhead execution module based on the registration offset to adaptively adjust the printing press registration.

[0007] Optionally, an initial image of the printing press and corresponding real-time printing speed data are acquired while the high-speed rotary printing press is in motion; based on the real-time printing speed data, a preset dynamic point spread function model is matched to construct a motion blur degradation matrix; the initial image is then subjected to Wiener filtering restoration processing using the motion blur degradation matrix to generate the image with motion blur eliminated.

[0008] Optionally, based on the image, a linear region extending along the printing paper feed direction and having a continuous high grayscale value is extracted as the main feature region of the reference line; connected component growth detection is performed outward from the edge of the main feature region of the reference line; the region where the growth detection shows a spatial intersection with the alignment mark and a continuous grayscale gradient without discontinuities is determined as the connected fusion region.

[0009] Optionally, based on the connected fusion region, a one-dimensional grayscale profile curve is extracted along the normal direction pointing to the alignment mark from the reference line feature; the one-dimensional grayscale profile curve is differentiated to obtain the grayscale gradient change rate curve of ink diffusion energy attenuation; based on the grayscale gradient change rate curve, the local extreme points of the grayscale gradient change rate are captured as the physical boundary of the alignment mark.

[0010] Optionally, pixels outside the physical boundary and located within the connected fusion region are set as pixels to be processed; the pixel values ​​of the pixels to be processed are replaced with background primary color values ​​to generate the reconstructed independent marker image.

[0011] Optionally, in an area outside the connected fusion region and uncontaminated by ink, multiple background reference pixels are extracted; the median grayscale value of the multiple background reference pixels is calculated as the background primary color value; and the pixel value of the pixel to be processed is replaced with the background primary color value to generate the independent marker image.

[0012] Optionally, based on the independent marked image, edge gradient distribution features are extracted; by continuously fitting and deducing the edge gradient distribution features, the continuous centroid coordinates of the alignment marks corresponding to the color plates are located respectively; based on the pixel energy features of the alignment marks, the alignment mark with the highest energy value is selected as the origin mark, and the continuous centroid coordinates of the origin mark are set as the origin of the reference coordinate system; based on the origin of the reference coordinate system, the physical offset distance of the continuous centroid coordinates of the alignment marks corresponding to other color plates relative to the origin of the reference coordinate system in the vertical paper feeding direction and along the paper feeding direction is calculated, and used as the overprinting offset.

[0013] Optionally, based on the overprinting offset, an initial compensation step size is calculated in conjunction with real-time operating parameters of printing speed and paper tension. The overprinting offset includes a vertical paper feed direction offset component and a paper feed direction offset component. The vertical paper feed direction offset component corresponds to the calculated mechanical transmission compensation step size, and the paper feed direction offset component corresponds to the calculated control board signal compensation step size. The control command is generated based on the initial compensation step size. Based on the execution response delay time after the control command is issued, a command silence suppression period is set. The execution response delay time is determined comprehensively based on the mechanical transmission delay time corresponding to the mechanical transmission compensation step size and the electrical signal delay time corresponding to the control board signal compensation step size. During the command silence suppression period, the generation of the next round of control commands is blocked until re-sampling verification confirms that the overprinting offset has converged to a preset convergence threshold.

[0014] Optionally, after the instruction silence period expires, a preset number of re-acquired images are continuously acquired, and the real-time overprint offset corresponding to the re-acquired images is calculated respectively; the temporal variance of the real-time overprint offset is calculated; when the temporal variance is less than a preset stability threshold and the mean of the real-time overprint offset is within a preset convergence threshold range, convergence is determined, and the blocking of the generation of the next round of control instructions is released.

[0015] Secondly, this application also provides an adaptive control system for a high-speed rotary printing press based on dynamic monitoring, used to execute the adaptive control method for a high-speed rotary printing press based on dynamic monitoring as described in the first aspect. The adaptive control system for a high-speed rotary printing press based on dynamic monitoring includes: a format image acquisition module for acquiring a format image during the high-speed printing process, the format image containing reference line features and alignment marks; a connected fusion region identification module for identifying, based on the format image, a connected fusion region formed by the dynamic diffusion of ink between the reference line features and the alignment marks; an anti-diffusion feature reconstruction module for reconstructing anti-diffusion features of the alignment marks based on the connected fusion region to obtain an independent mark image; an offset calculation module for calculating the registration offset between each color plate based on the independent mark image; and a registration adjustment module for generating and issuing control commands to the printhead execution module to perform adaptive registration adjustment of the printing press based on the registration offset.

[0016] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0017] By acquiring a format image during the high-speed printing process, which includes reference line features and alignment marks, and based on this image, identifying the connected and fused regions formed by the dynamic diffusion of ink between the reference line features and the alignment marks, and reconstructing the alignment marks using anti-diffusion features based on these connected and fused regions to obtain independent mark images, and calculating the registration offset between each color plate based on these independent mark images, and generating and issuing control commands to the printhead execution module to adaptively adjust the printing press registration, the system significantly improves registration accuracy and printing quality during high-speed printing by actively identifying connected and fused regions and reconstructing the alignment marks using anti-diffusion features.

[0018] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the adaptive control method for a high-speed rotary printing press based on dynamic monitoring, as described in this application.

[0021] Figure 2 This is the adaptive convergence curve of the total registration offset of each color plate in the adaptive control method of high-speed rotary printing press based on dynamic monitoring in this application.

[0022] Figure 3 This is a schematic diagram of the adaptive control system for a high-speed rotary printing press based on dynamic monitoring, as described in this application.

[0023] Figure labeling: 11 image acquisition module, 12 connected fusion region identification module, 13 anti-diffusion feature reconstruction module, 14 offset calculation module, 15 overprint adjustment module. Detailed Implementation

[0024] This application provides an adaptive control method and system for high-speed rotary printing presses based on dynamic monitoring. This solves the technical problem in existing technologies where dynamic ink diffusion in high-speed printing environments causes distortion of registration marks, leading to misregistration and further affecting print quality. By actively identifying connected and fused regions and reconstructing the anti-diffusion features of registration marks to restore independent mark images, the system calculates the actual physical misregistration between color plates based on the reconstructed clean mark images. Control commands are then generated and issued, ensuring the adaptive control system consistently corrects misregistration based on the actual error, thereby significantly improving registration accuracy and print quality in high-speed printing processes.

[0025] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0026] Example 1, please refer to the appendix. Figure 1This application provides an adaptive control method for a high-speed rotary printing press based on dynamic monitoring. The method is applied to an adaptive control system for a high-speed rotary printing press based on dynamic monitoring. The method specifically includes the following steps:

[0027] Acquire a sheet image during the high-speed printing process, the sheet image containing reference line features and alignment marks.

[0028] Furthermore, this application also includes the following steps: acquiring an initial image of the printing press in motion and corresponding real-time printing speed data; based on the real-time printing speed data, matching a preset dynamic point spread function model to construct a motion blur degradation matrix; and using the motion blur degradation matrix to perform Wiener filtering restoration processing on the initial image to generate the image with motion blur eliminated.

[0029] Specifically, during normal high-speed operation of the printing press, the Eagle Eye Detection System acquires the initial image of the current frame. This system consists of multiple high-definition cameras evenly distributed below the printhead and on both sides of the printing surface. The camera optical axes are perpendicular to the printing surface, and a ring light source is coaxially mounted at the front of the lens to provide uniform, shadowless illumination. It acquires real-time images of the printhead status, the printing surface, the printhead junction area, and the finished print during the printing process. The overprinting detection module has a detection range of 24mm*20mm, a normal detection speed of 150m / min, an optical resolution of 10μm, and an overprinting measurement accuracy of 20μm. It performs image acquisition and overprinting offset calculations under normal conditions.

[0030] A rotary encoder is installed on the main drive shaft of the printing press. The encoder pulse signal is synchronized with the camera exposure trigger signal. When the encoder pulse count reaches a preset threshold, the camera is triggered to expose and capture a frame image. At the same time, the current encoder pulse frequency is read and converted into real-time printing speed data to ensure that the image data and speed data are strictly aligned in time.

[0031] The real-time printing speed data is converted into physical displacement distance within the exposure time. Combined with the camera's physical pixel equivalent, the pixel length spanned by the motion blur in the image space under the current conditions is calculated. Using this pixel length as an index, a preset dynamic point spread function model library is accessed. If a PSF kernel precisely corresponding to this length exists in the dynamic point spread function model library, it is directly retrieved; otherwise, the kernels of two adjacent speed levels are taken, and linear interpolation is performed according to the relative position of the actual speed between them to generate an adaptive kernel specific to the current speed. The PSF kernel is a two-dimensional matrix whose size covers the range of blur length. Each element along the central axis of the paper feed direction is assigned the reciprocal of the blur length, and the remaining positions are set to 0 to ensure energy conservation.

[0032] The preset dynamic point spread function (PSF) model is a library of mathematical convolution kernels pre-stored in the control system's memory. Each PSF kernel in this library corresponds to a specific speed level or fuzzy length. Since the printing press's motion direction is constant under normal operating conditions, along the paper feed direction, it is typically a one-dimensional horizontal linear kernel, with the fuzzy length as its core parameter. The significance of the preset model is to avoid the computational delay caused by online real-time PSF estimation.

[0033] A Fast Fourier Transform (FPF) is performed on the spatial domain kernel matrix. Before the transform, to avoid frequency domain aliasing, the kernel matrix is ​​cyclically shifted to move the center zero to the origin, and zero-padding is used to expand the kernel matrix size to match the original image size. After a two-dimensional Discrete Fourier Transform (DFT), a frequency domain matrix with the same resolution as the image and each element being a complex number is output; this is the motion blur degradation matrix at the current moment. The motion blur degradation matrix is ​​the complex transfer function matrix obtained by transforming the selected spatial domain PSF kernel to the frequency domain using a DFT. In image restoration theory, the complex transfer function matrix describes how the frequency components of a sharp image are attenuated and phase-shifted under motion.

[0034] Based on the motion blur degradation matrix and combined with preset noise-to-signal power ratio parameters, a Wiener filter restoration transfer function is constructed. That is, the pre-calibrated noise-to-signal power ratio parameters stored in the controller are read in, and element-wise operations are performed on the motion blur degradation matrix: the square of the amplitude at each frequency point is calculated, and the Wiener filter formula W(u,v)=conj(H(u,v)) / (|H(u,v)| is applied. 2 The restoration transfer function is constructed using the regularization term +K, where H(u,v) is the motion blur degradation matrix; K is a preset signal-to-noise ratio parameter, such as 0.01; and conj(H(u,v)) represents taking the complex conjugate of the degradation matrix. For frequency points where the amplitude of H(u,v) is close to zero, i.e., notch zeros generated by motion blur in the frequency domain, the division-by-zero error is avoided by the regularization term +K in the denominator, thus achieving controllable compensation for lost frequency components.

[0035] The initial image is also transformed to the frequency domain, i.e., a two-dimensional discrete Fourier transform is performed to obtain its frequency domain representation G(u,v). The frequency domain representation G(u,v) is then multiplied element-wise with the Wiener filter transfer function W(u,v) in the frequency domain to obtain a frequency domain estimate of the restored image. A two-dimensional discrete inverse Fourier transform is then performed, and the real part of the result is taken. After grayscale truncation and normalization, a restored image with motion blur largely eliminated is output, containing clear, sharp reference line features and alignment marks. The Wiener filter restoration process comprehensively considers the inverse transformation of the degradation matrix and the image signal-to-noise ratio. While eliminating motion blur, it effectively suppresses high-frequency noise that is easily amplified by simple inverse filtering, thus solving for the minimum mean square error estimate of the original clear image. Reference line features, also known as maintenance lines, are linear areas with continuous high gray values ​​that extend along the printing paper feed direction; alignment marks, also known as test marks, are specific shaped color blocks printed on the edges of each color plate, such as cross lines or rectangular blocks.

[0036] By constructing a Wiener filter that is strictly matched to the real-time speed, the linear blurring that is unavoidable in high-speed motion is removed from the image, allowing the edge gradients of reference lines and alignment marks in the output image to be restored to sharpness. Since the PSF model is dynamically matched based on real-time speed data rather than being fixed, it can still adaptively adjust the blur kernel even during the acceleration and deceleration phases of the printing press, ensuring image consistency across the entire speed range.

[0037] Based on the image, the connected and fused region formed by the dynamic diffusion of ink between the reference line features and the alignment mark is identified.

[0038] Furthermore, this application also includes the following steps: based on the image, extracting a linear region extending along the printing paper feed direction and having a continuous high grayscale value as the main region of the reference line feature; performing connected component growth detection outward from the edge of the main region of the reference line feature; determining the region where the growth detection shows a spatial intersection with the alignment mark and a continuous grayscale gradient without breaks as the connected fusion region.

[0039] Specifically, a global grayscale statistical analysis is performed on the restored image. An adaptive binarization threshold is set to segment the image into foreground and background binary images, where the foreground is the ink-covered area and the background is the substrate area. Morphological opening operations are performed on the binary images along the paper feed direction to remove isolated noise and connect broken line segments. Then, continuous high-grayscale connected regions extending horizontally beyond a preset length threshold are extracted as the main feature region of the reference line, and their edge contour coordinates are recorded. The preset length threshold is the minimum length threshold used to distinguish true reference lines from short connected regions formed by random ink dots or noise. Only regions continuously extending horizontally beyond the preset length threshold (e.g., 200 pixels) are identified as main reference lines. This reduces the false detection rate, but may result in missed detections when reference lines are broken due to uneven ink distribution or wear; therefore, the preset length threshold is increased. Improved robustness allows for the identification of shorter or broken reference lines, but may misidentify noise or accidentally formed long ink marks as reference lines; therefore, the preset length threshold is decreased.

[0040] Because the mechanical motion of the printing press is constant along this direction, the dynamic diffusion of ink also mainly produces a trailing effect along this direction. Therefore, the connection and fusion of the reference line and the mark mainly occurs along the paper feed direction. A continuous linear region with high grayscale values ​​is a set of pixels extending along the paper feed direction. Its grayscale value is significantly higher than the surrounding background area, and it is spatially distributed in a long, continuous strip shape without obvious breaks or discontinuities. The main feature area of ​​the reference line does not include the edge transition zone of the reference line, but refers to its main body with stable grayscale and complete shape.

[0041] Starting from the outer edge pixels of the main feature region of the reference line, a line-by-line scan is performed perpendicular to the paper feed direction, i.e., the normal direction. For each row of pixels, the process moves outward pixel by pixel along the normal direction, recording the grayscale value of the current pixel and calculating the grayscale gradient magnitude between it and the previous pixel. When the grayscale value of a pixel drops to near the background grayscale level, i.e., the grayscale gradient is below a preset threshold, growth in that direction is stopped, thus obtaining the growth termination point. This operation is repeated for all edge pixels to construct the complete region growing outward from the reference line. When growing connected components outward from the edge of the reference line, when the grayscale gradient magnitude between adjacent pixels is below a preset threshold, it is considered that the region has left the ink diffusion influence area, i.e., reached the background, and growth stops. Relaxing the growth stopping condition will cause the connected components to expand outward more, possibly including background noise areas in the connected fusion region, resulting in a larger subsequent reconstruction range. In this case, the preset threshold is increased. Tightening the growth stopping condition will reduce the connected component expansion range, possibly missing connected fusion regions with less ink and milder diffusion. In this case, the preset threshold is decreased.

[0042] Check if the grown region spatially overlaps with the location of the alignment marker in the image. If overlap exists, extract the grayscale gradient sequence along the path from the reference line edge to the alignment marker edge, and determine if there are abrupt gradient changes or discontinuities in the grayscale gradient sequence, i.e., a sudden increase or irregular jump in grayscale value from high to low. If the grayscale gradient remains monotonically and continuously changing along the entire path without significant discontinuities or abrupt changes, then the grown region is determined to be a connected and fused region formed by ink diffusion, and its boundary coordinate range is recorded. Output the location information and pixel set of the connected and fused region.

[0043] A continuous, unbroken grayscale gradient is characterized by a continuous transition in the rate of change of grayscale values ​​along the pixel path from the edge of the main reference area to the alignment marker, without abrupt changes or breaks. A continuous grayscale gradient means that the connection between the two is a natural grayscale transition band formed by ink diffusion, rather than an accidental connection caused by noise or image artifacts. When the growing region simultaneously satisfies both the conditions of spatial intersection with the marker and continuous, unbroken grayscale gradient, it can be determined as a truly connected and fused region, a contaminated area suitable for anti-diffusion reconstruction.

[0044] By accurately extracting the main area of ​​the reference line and performing directional growth detection based on grayscale gradients, the connected and merged regions caused by ink diffusion are accurately separated from the complex image background, avoiding indiscriminate full-image processing. Through the strict criterion of continuous and unbroken grayscale gradients, the system effectively distinguishes connected regions formed by the natural transition of ink diffusion from the accidental spatial proximity of two independent color blocks caused by misregistration.

[0045] Based on the connected fusion region, the alignment markers are reconstructed using anti-diffusion features to obtain independent marker images.

[0046] Furthermore, this application also includes the following steps: based on the connected fusion region, extracting a one-dimensional grayscale profile curve along the normal direction pointing to the alignment mark from the reference line feature; performing derivative processing on the one-dimensional grayscale profile curve to obtain a grayscale gradient change rate curve of ink diffusion energy attenuation; based on the grayscale gradient change rate curve, capturing the local extreme points of the grayscale gradient change rate as the physical boundary of the alignment mark.

[0047] Furthermore, this application also includes the following steps: setting the pixels outside the physical boundary and located within the connected fusion region as pixels to be processed; replacing the pixel values ​​of the pixels to be processed with background primary color values ​​to generate the reconstructed independent marker image.

[0048] Furthermore, this application also includes the following steps: extracting multiple background reference pixels in an area outside the connected fusion region that is not contaminated by ink; calculating the median grayscale value of the multiple background reference pixels as the background primary color value; and replacing the pixel value of the pixel to be processed with the background primary color value to generate the independent marker image.

[0049] Specifically, based on the accurate positioning of the connected and fused regions, the physical boundaries of the alignment markers are accurately located through mathematical analysis of the one-dimensional grayscale profile. Then, all the diffusion artifact pixels are replaced with the background color, and an independent marker image is output, which is a pure binary or grayscale image containing only the real ink area of ​​the alignment marker itself.

[0050] Based on the boundary coordinates of the connected and blended region, the spatial correspondence between the reference line and the alignment mark is determined. For each horizontal position (X-coordinate) within the connected and blended region, starting from the inner edge of the main body of the reference line, grayscale values ​​are read pixel by pixel along the normal direction pointing towards the alignment mark (vertically downwards) until the alignment mark is reached or the background region is reached, thus constructing a one-dimensional grayscale profile curve. The one-dimensional grayscale profile curve is a sequence of pixel grayscale values ​​along the normal direction at a fixed horizontal position, starting from the main body of the reference line, passing through the connected and blended region, extending to the alignment mark or reaching the background region, reflecting the grayscale gradation process caused by ink diffusion.

[0051] The grayscale gradient rate of change curve is obtained by differentiating the one-dimensional grayscale profile curve. Within the reference line ink area, the grayscale value remains stable at a high level, and the gradient is close to 0. In the ink diffusion transition area, the grayscale value continuously decreases, and the gradient is negative. Upon reaching the alignment mark, the grayscale value changes from decreasing to increasing, and the gradient changes from negative to positive, reaching a positive peak. The grayscale gradient rate of change curve is the gradient curve obtained by differentiating the one-dimensional grayscale profile curve, reflecting the rate of change of grayscale value per pixel movement along the normal direction. In the ink diffusion region, the grayscale value gradually transitions from high to low, with a negative gradient and a gentle change. Upon reaching the alignment mark at the diffusion front, the grayscale value increases again from low, and the gradient changes from negative to positive, forming a characteristic valley-to-peak shape.

[0052] Search for local extrema on the gray-level gradient rate of change curve, capturing the positive maximum immediately following the negative minimum. Record the coordinates of the normal direction corresponding to the positive maximum as the physical boundary of the alignment mark at that horizontal position. Repeat the above operation for all horizontal positions within the connected fusion region to obtain the complete physical boundary outline of the alignment mark. Local extrema are points on the gray-level gradient rate of change curve where the gray-level gradient value reaches a local maximum or minimum within a certain neighborhood. Focus on the negative local minimum and the positive local maximum that appear during the process of the gradient gradually decreasing from negative to positive and then rapidly rising back to positive. The negative minimum corresponds to the position where the gray level decreases most sharply when the ink spreads from the reference line to the background, i.e., the leading edge of the ink; the positive maximum corresponds to the position where the gray level increases most sharply when the ink spreads from the background to the alignment mark, i.e., the physical edge of the mark; the area between the two is the effective diffusion transition zone of the ink.

[0053] Based on the physical boundaries and the spatial range of the identified connected fusion regions, spatial logic is performed pixel-by-pixel in the image coordinate system. For each pixel within the connected fusion region, it is checked whether the pixel's position in the normal direction is outside the physical boundary, i.e., away from the alignment mark core. If this condition is met (i.e., located outside the physical boundary and within the connected fusion region), the pixel is marked as a pixel to be processed, and its coordinates and current grayscale value are recorded. If this condition is not met (i.e., the pixel is located inside the physical boundary and belongs to the true ink area of ​​the alignment mark), it is retained without any processing. After completing the pixel filtering of the entire region, the background primary color value is estimated.

[0054] Multiple sampling windows are defined around the connected and fused regions, maintaining a sufficient safe distance from the boundaries of the connected and fused regions to avoid indirect influence from diffused ink blots; other ink blot areas that may exist in the image, such as adjacent color stops and text, are also avoided. Within each sampling window, the grayscale values ​​of hundreds of pixels are extracted to form a background reference sample set.

[0055] The median grayscale value of multiple background reference pixels is calculated; that is, the median value is taken after sorting the multiple background reference pixels and used as the background primary color value. All pixels already marked for processing are iterated over, and their original grayscale values ​​are uniformly replaced with this background primary color value. After the replacement, the portions of the original connected fusion region that belong to diffusion artifacts all become the background color. A continuous background color separating band appears between the reference line and the alignment mark, and the two are no longer connected, thus generating an independent marker image. In the independent marker image, the reference line and the alignment mark are completely separated by the background color, and there is no longer any connection or adhesion. The alignment mark is presented in an independent form, and its geometric centroid is determined only by the actual ink of the mark itself.

[0056] By mathematically differentiating the one-dimensional grayscale profile along the normal direction and capturing local extrema, the true physical boundary of the alignment mark is located with sub-pixel precision. Using the grayscale median as the background base color significantly improves noise resistance. The alignment mark in the independent mark image no longer contains any diffusion artifacts from the reference line; its geometric centroid is determined solely by the actual ink on the mark itself. This fundamentally eliminates the core problem of detection signal distortion caused by ink diffusion in existing technologies, ensuring that the calculated overprint offset truly reflects the mechanical displacement between the rollers.

[0057] Based on the independent marked image, the overprinting offset between each color plate is calculated.

[0058] Furthermore, this application also includes the following steps: extracting edge gradient distribution features based on the independent marker image; locating the continuous centroid coordinates of the alignment marks corresponding to the color plates by performing continuous fitting and deduction on the edge gradient distribution features; selecting the alignment mark with the highest energy value as the origin mark based on the pixel energy features of the alignment marks, and setting the continuous centroid coordinates of the origin mark as the origin of the reference coordinate system; calculating the physical offset distance of the continuous centroid coordinates of the alignment marks corresponding to other color plates relative to the origin of the reference coordinate system in the vertical paper feeding direction and along the paper feeding direction, based on the origin of the reference coordinate system, as the overprinting offset.

[0059] Specifically, gradient magnitude is calculated for the local region containing each alignment marker in the independently labeled image. The Sobel operator is used to calculate the first-order derivative of each pixel in the horizontal and vertical directions, obtaining the gradient magnitude and gradient direction angle for each pixel. Pixels with gradient magnitudes greater than a preset edge threshold are marked as edge candidates, and their sub-pixel position coordinates are recorded. The preset edge threshold is a critical gradient magnitude value used to distinguish effective edges from background noise / non-edge regions. When the gradient magnitude of a pixel exceeds the preset edge threshold, the pixel is marked as an edge candidate; otherwise, it is not considered an edge. The gradient magnitudes of all pixels within the local region are statistically analyzed and sorted by magnitude; the 85th to 90th percentile of the gradient magnitude distribution is taken as the preset edge threshold.

[0060] Each alignment mark is processed independently. For a given color mark, the set of pixel coordinates of all candidate edge points is processed using a least-squares elliptic fitting method for continuity. The elliptic fitting model is preferred because, under normal printing conditions, alignment marks are usually designed as circles, crosshairs, or rectangles, and after imaging, due to perspective or slight deformation, they often appear as approximate elliptical shapes. The elliptical model has strong shape adaptability and ease of analytical solution. During the fitting process, an objective function is constructed, which is the sum of squared algebraic distances from each edge point to the boundary of the fitted ellipse. By solving for the minimum value of this objective function, the standard parameters of the ellipse equation are calculated, including the coordinates of the ellipse center, the major semi-axis, the minor semi-axis, and the rotation angle. After completing the continuous fitting, a mathematical description of the mark contour is obtained, filtering out the influence of isolated noise points that may exist among the discrete edge points.

[0061] Based on the fitted continuous contour equation, the internal region of the marker is determined, which is the set of all pixels enclosed by extending along the major and minor axes to the contour boundary, with the center of the fitted ellipse as the reference. The grayscale values ​​of these internal pixels are read, and a grayscale-weighted centroid calculation is performed. Using the grayscale value of each pixel as the weight of its coordinates, the weighted sum of the abscissa and ordinate of all pixels within the marker region is calculated, and then divided by the total weight (the sum of all pixel grayscale values) to obtain the continuous centroid coordinates of the marker. The weighted centroid calculation integrates the grayscale distribution information within the marker, which has a certain suppressive effect on shape disturbances caused by uneven ink distribution or slight changes in illumination at the marker edges, resulting in more robust positioning results. The above complete process is repeated for all alignment markers corresponding to different color versions, outputting their respective continuous centroid coordinates with sub-pixel precision. The continuous centroid coordinates are the geometric center coordinates of the marker with sub-pixel precision, obtained by performing a grayscale-weighted average calculation on all pixels within the marker's internal region after the continuous fitting and derivation are completed. They are typically represented as two-dimensional coordinates.

[0062] Calculate the pixel energy feature value of each alignment mark, which is the sum of the grayscale values ​​of all pixels within the marked area. Iterate through the energy values ​​of all marks, identifying the mark with the highest energy value as the origin mark, and setting its continuous centroid coordinates as the origin of the reference coordinate system. Process the remaining color plate marks one by one, excluding the origin mark. For each alignment mark corresponding to another color plate, perform a difference operation between its continuous centroid coordinates and the continuous centroid coordinates of the origin mark in the reference coordinate system to obtain the pixel deviation of each mark in the image coordinate system. Multiply the pixel deviation by the pre-calibrated physical pixel equivalent and apply it independently along the paper feed direction and perpendicular to the paper feed direction to obtain the physical offset distance of each color plate in these directions. The physical pixel equivalent is the actual physical spatial size corresponding to a single pixel in the image, which is the actual distance between the centers of two adjacent pixels in the camera's field of view. On a calibration board with known physical dimensions, count the pixel distances between feature points on the calibration board, and divide the physical distance by the pixel distance to obtain the physical pixel equivalent. According to the preset sign convention, such as positive along the paper feed direction and positive perpendicular to the paper feed direction, the direction of offset is determined, and the sign is retained in the output result. (Color plate number, Δx) mm ,Δy mm The format organizes the overprinting offset data of each color plate into a structured data packet, where the plus or minus sign clearly indicates the offset direction: +Δx mm Indicates offset along the paper feed direction, -Δx mm Indicates a deviation in the reverse paper feed direction; +Δy mm This indicates a shift along the vertical paper feed direction towards the positive direction of the printhead array, -Δy mm This indicates a shift in the negative direction of the printhead array. The paper feed direction is the X-axis of the image coordinate system, corresponding to the printing paper feed direction, i.e., the direction in which the paper moves at high speed below the printhead. The physical offset distance along the paper feed direction reflects the actual registration deviation of the color plate in that direction. The direction perpendicular to the paper feed direction is the Y-axis of the image coordinate system, perpendicular to the printing paper feed direction, i.e., the paper width direction. The physical offset distance perpendicular to the paper feed direction reflects the actual registration deviation of the color plate in the direction perpendicular to the paper feed direction.

[0063] By introducing edge gradient continuous fitting technology, the centroid positioning accuracy was improved from the whole pixel level to the sub-pixel level, achieving micrometer-level measurement resolution. Selecting the marker with the highest energy as the origin effectively avoided the positioning drift problem caused by weak signals in light-colored ink layers.

[0064] Based on the overprint offset, control commands are generated and sent to the printhead execution module to perform adaptive overprint adjustment of the printing press.

[0065] Furthermore, this application also includes the following steps: calculating an initial compensation step size based on the registration offset and real-time operating parameters of printing speed and paper tension, wherein the registration offset includes a vertical paper feed direction offset component and a paper feed direction offset component, the vertical paper feed direction offset component corresponds to the calculated mechanical transmission compensation step size, and the paper feed direction offset component corresponds to the calculated control board signal compensation step size; generating the control command based on the initial compensation step size; setting a command silence suppression period based on the execution response delay time after the control command is issued, wherein the execution response delay time is determined comprehensively based on the mechanical transmission delay time corresponding to the mechanical transmission compensation step size and the electrical signal delay time corresponding to the control board signal compensation step size, blocking the generation of the next round of control commands during the command silence suppression period until the registration offset is verified to converge to a preset convergence threshold.

[0066] Furthermore, this application also includes the following steps: after the instruction silence suppression period expires, continuously acquire a preset number of re-acquired images of the same frame size, and calculate the real-time overprint offset corresponding to the re-acquired images of each frame size; calculate the temporal variance of the real-time overprint offset; when the temporal variance is less than a preset stability threshold and the mean of the real-time overprint offset is within a preset convergence threshold range, convergence is determined, and the blocking of the generation of the next round of control instructions is released.

[0067] Specifically, the registration offset data of each color plate is acquired and decomposed into offset components along the paper feed direction and offset components perpendicular to the paper feed direction. For the offset component along the paper feed direction, the printing speed in the current real-time operating parameters is read based on its magnitude and direction. The compensation along the paper feed direction is directly related to the speed; the higher the speed, the greater the paper feed distance per unit time, and the phase compensation equivalent of the ignition timing also changes accordingly. Combined with the preset control board signal compensation coefficient, the control board signal compensation step size is calculated. The control board signal compensation step size is ultimately output in the form of encoder pulse count or time delay value, used to adjust the trigger timing of the printhead ignition signal.

[0068] For the vertical paper feed offset component, the current printing speed and paper tension are simultaneously read. The mechanical transmission in the vertical paper feed direction is mainly affected by the load. Tension fluctuations affect the accuracy of the lead screw transmission and the force state of the printhead mount. Combining the preset mechanical transmission compensation coefficients, including speed adaptive gain and tension feedforward compensation, the mechanical transmission compensation step size is calculated to drive the servo motor to move the printhead mount along the vertical paper feed direction. The calculations of the two components are independent and performed in parallel.

[0069] Specifically, the registration offset data and corresponding color plate numbers for each color plate are obtained. For the reference color plate, the offset is (0,0), and no adjustment command needs to be generated. For each non-reference color plate with out-of-tolerance settings, a dual-axis parallel compensation step size calculation process is initiated. The printing speed v in the current real-time operating parameters is read, and the speed adaptive proportional gain K is calculated based on the current speed. p =K p0 ×(v0 / v) α K p0 The baseline proportional gain is calibrated using a step response experiment at a standard speed v0, and α is the speed correction exponent, such as 0.4. Subsequently, the speed adaptive proportional gain K... p Multiply by the offset component along the paper feed direction to obtain the physical equivalent compensation amount. Read the pre-stored encoder equivalent, the physical paper feed distance corresponding to each encoder pulse, and convert the physical compensation amount into the encoder pulse number, i.e., the ratio of the physical equivalent compensation amount to the encoder equivalent, and round it to the nearest integer. The compensation step size along the paper feed direction is marked with a positive or negative sign; a positive sign indicates advance ignition, i.e., the ink droplet moves forward, and a negative sign indicates delayed ignition, i.e., the ink droplet moves backward.

[0070] Simultaneously read the current printing speed and paper tension, calculate the tension deviation (obtained by subtracting the tension reference value from the current paper tension), and calculate the speed adaptive proportional gain K based on the current speed. p The same speed adaptive proportional gain is used along the paper feed direction, and multiplied by the offset component perpendicular to the paper feed direction to obtain the proportional term output. The pre-stored tension feedforward coefficient, i.e., the displacement to be compensated for every 1N change in tension, is read, and the tension feedforward compensation term is calculated as: Tension Feedforward Coefficient × Tension Deviation. The proportional term and the tension feedforward term are superimposed to obtain the mechanical transmission compensation step size, which is rounded to the minimum resolution of the actuator, typically 0.001mm. The compensation step size perpendicular to the paper feed direction is signified; a positive sign indicates movement in the positive direction of the printhead array, and a negative sign indicates movement in the negative direction of the printhead array.

[0071] The initial compensation step size, target color plate number, execution timestamp, and other information, including compensation parameters along the paper feed direction and perpendicular to the paper feed direction, are encapsulated into control commands and sent to the printhead execution module via industrial Ethernet. The printhead execution module is electrically connected to the printhead assembly of the high-speed rotary printing press, receives control commands from the adaptive control module, and performs operations such as nozzle filling, overprint adjustment, and printhead misalignment adjustment. After the command is issued, a command silence period is set based on the execution response delay time, i.e., the time window during which the generation of the next round of control commands is prohibited.

[0072] The length of the command silence period is set based on the execution response delay time. It is the time delay between the issuance of the control command and the actual completion of the action by the actuator, resulting in a change in the overprint offset. It is determined by the mechanical transmission delay time corresponding to the mechanical transmission compensation step size and the electrical signal delay time corresponding to the control board signal compensation step size. Since the mechanical transmission delay is much greater than the electrical signal delay, the overall execution response delay time is mainly determined by the mechanical transmission delay. The mechanical transmission delay time is calibrated during the equipment debugging phase through a step response experiment. A position command with a fixed step size is issued to the servo driver, and the time from the issuance of the command to the actual position change of the printhead mount reaching 90% of the target value is recorded; this is the mechanical transmission delay time. The electrical signal delay time is provided by the control board hardware specifications and is usually a fixed empirical value, such as 1 μs. Multiplying the overall execution response delay time by a safety factor k, typically between 1.5 and 2, yields the command silence period. The selection of the safety factor considers the response time fluctuations of the mechanical transmission system under different operating conditions and the mutual interference between the actuators of each color plate during concurrent control of multiple color plates.

[0073] During the instruction silence period, the generation logic of all new rounds of control instructions is blocked, waiting for the actuator to complete the action response. The blocked objects are all control instructions in both the paper feed direction and the perpendicular paper feed direction. After the silence period expires, the block is lifted, but no new instructions are generated immediately. Instead, the convergence verification phase begins. A preset number of re-acquisition images are continuously acquired, such as 10 frames. For each frame, following the complete previous process, the corresponding real-time overprint offset sequence is calculated, and its arithmetic mean and temporal variance are calculated. The temporal variance is compared with a preset stability threshold, and the absolute value of the mean is compared with a preset convergence threshold. If both conditions are met simultaneously, the control is considered to have successfully converged, the complete blockade on the generation of the next round of control instructions is lifted, the current round of adaptive overlay adjustment is completed, and the system enters the normal monitoring state to prepare to respond to the detection results of the next frame. If either condition is not met, the system is considered to have not yet converged, and the control instructions are generated again based on the latest offset. This cycle is repeated until the convergence conditions are met or the maximum number of control cycles is exceeded.

[0074] The preset stability threshold is the upper limit of the time series variance used to determine whether the system has stabilized. When the time series variance is less than this threshold, the system is considered to have entered a stable state without continuous oscillation. The preset convergence threshold is the error limit used to determine whether the overprinting offset has met the printing accuracy requirements. When the mean of the offset is within this threshold range, the overprinting accuracy is considered to have met the standard. The typical value range is 0.02 to 0.05 mm.

[0075] For example, as shown in the appendix Figure 2As shown, taking four-color printing on a high-speed rotary printing press as an example, the initial total registration offsets of the cyan, magenta, and yellow plates are approximately 56μm, 61μm, and 58μm, respectively. After adaptive adjustment, the total offset of each plate has dropped below 5μm in the 7th to 8th adjustment cycles, entering the convergence threshold region, i.e., the gray shaded area. By the 11th cycle, it stabilizes within the range of 2 to 3μm. Considering the silent suppression mechanism of mechanical transmission delay, effective convergence of registration offset can be achieved. Assuming a printing speed of 180m / min, paper tension of 280N / m, camera resolution of 4096×1 pixels, pixel size of 7.04μm, and a physical width corresponding to the camera's field of view of 28.83mm, the physical pixel equivalent is 0.00704mm / pixel, the image acquisition frame rate is 100fps, i.e., the sampling interval is 10.0ms; the registration accuracy requires the total offset to converge to within 5.0μm. After anti-diffusion feature reconstruction and continuous centroid positioning, the total offset of the cyan plate was measured to be 56.8 μm, the total offset of the magenta plate to be 61.2 μm, and the total offset of the yellow plate to be 58.5 μm. The current operating parameters were read: printing speed 180.0 m / min, paper tension 280.0 N / m. The initial compensation step size was calculated to be 1.00276, resulting in a compensation step size of 56.96 μm for the cyan plate, 61.37 μm for the magenta plate, and 58.66 μm for the yellow plate. Control commands were issued, with a set command silence period of 50 ms, a mechanical transmission delay of 200 ms, and a stabilization time of 250 ms. During this silence period, the generation of the next round of control commands was blocked. Periods 0 to 3 represent the rapid descent phase, with the offset decreasing from 60 μm to 20 μm; periods 4 to 6 represent the deceleration phase, with the offset decreasing from 20 μm to 8 μm; periods 7 to 11 represent the convergence and stabilization phase, where the offset enters the 5 μm threshold and eventually stabilizes at 2 to 3 μm. After the 10th period, five frames of resampled images were continuously acquired, with temporal variance ranging from 0.128 to 0.189 μm. 2 Within the range, all are less than the preset stability threshold of 4μm. 2 The mean value ranges from 2.89 to 3.18 μm, which is within the preset convergence threshold range of [-5,5] μm. All three color plates simultaneously meet both conditions, thus convergence is determined and the silencing suppression is lifted.

[0076] The detected registration offset is translated into specific actions, enabling fully automatic adaptive control of registration accuracy on high-speed rotary printing presses. The data storage module stores original print files, acquired print image data, system operating parameters, and adjustment records, facilitating subsequent traceability and parameter optimization. The control command generation and execution mechanism possesses extremely high dynamic response capabilities and positioning accuracy, effectively eliminating accumulated errors during high-speed operation.

[0077] In summary, the adaptive control method for high-speed rotary printing presses based on dynamic monitoring provided in this application has the following technical effects: It acquires a web-format image during the high-speed printing process, the web-format image containing reference line features and alignment marks; based on the web-format image, it identifies the connected and fused regions formed by the dynamic diffusion of ink between the reference line features and the alignment marks; based on the connected and fused regions, it reconstructs the alignment marks using anti-diffusion features to obtain independent mark images; based on the independent mark images, it calculates the registration offset between each color plate; based on the registration offset, it generates and sends control commands to the printhead execution module to adaptively adjust the printing press for registration. In other words, by actively identifying connected and fused regions and reconstructing the alignment marks using anti-diffusion features to restore independent mark images, and calculating the actual physical registration offset between each color plate based on the reconstructed clean mark images, it generates and sends control commands, enabling the adaptive control system to always accurately correct errors based on the actual error, thereby significantly improving the registration accuracy and printing quality during high-speed printing.

[0078] Example 2: Based on the same inventive concept as the adaptive control method for high-speed rotary printing presses based on dynamic monitoring in Example 1, this application also provides an adaptive control system for high-speed rotary printing presses based on dynamic monitoring. Please refer to the appendix. Figure 3 The adaptive control system for a high-speed rotary printing press based on dynamic monitoring includes: a format image acquisition module 11, used to acquire format images during the high-speed printing process, the format images containing reference line features and alignment marks; a connected fusion region identification module 12, used to identify the connected fusion region formed by the dynamic diffusion of ink between the reference line features and the alignment marks based on the format images; an anti-diffusion feature reconstruction module 13, used to reconstruct the anti-diffusion features of the alignment marks based on the connected fusion region to obtain independent mark images; an offset calculation module 14, used to calculate the overprinting offset between each color plate based on the independent mark images; and an overprinting adjustment module 15, used to generate and send control commands to the printhead execution module to perform adaptive overprinting adjustment of the printing press based on the overprinting offset.

[0079] Furthermore, the image acquisition module 11 in the adaptive control system for a high-speed rotary printing press based on dynamic monitoring is also used to: acquire an initial image of the high-speed rotary printing press in motion and the corresponding real-time printing speed data; based on the real-time printing speed data, match a preset dynamic point spread function model to construct a motion blur degradation matrix; and use the motion blur degradation matrix to perform Wiener filtering restoration processing on the initial image to generate the image with motion blur eliminated.

[0080] Furthermore, the connected fusion region identification module 12 in the adaptive control system of the high-speed rotary printing press based on dynamic monitoring is also used to: extract a linear region extending along the printing paper feed direction and having a continuous high gray value based on the image, as the main feature region of the reference line; perform connected domain growth detection outward from the edge of the main feature region of the reference line; and determine the region that has a spatial intersection with the alignment mark and has a continuous gray gradient without discontinuity as the connected fusion region.

[0081] Furthermore, the anti-diffusion feature reconstruction module 13 in the adaptive control system for high-speed rotary printing presses based on dynamic monitoring is also used to: extract a one-dimensional grayscale profile curve based on the connected fusion region, along the normal direction pointing to the alignment mark from the reference line feature; perform derivative processing on the one-dimensional grayscale profile curve to obtain the grayscale gradient change rate curve of ink diffusion energy attenuation; and capture the local extreme points of the grayscale gradient change rate based on the grayscale gradient change rate curve as the physical boundary of the alignment mark.

[0082] Furthermore, the anti-diffusion feature reconstruction module 13 in the high-speed rotary printing press adaptive control system based on dynamic monitoring is also used to: set the pixels outside the physical boundary and located in the connected fusion region as pixels to be processed; replace the pixel value of the pixels to be processed with the background primary color value to generate the reconstructed independent marker image.

[0083] Furthermore, the anti-diffusion feature reconstruction module 13 in the high-speed rotary printing press adaptive control system based on dynamic monitoring is also used to: extract multiple background reference pixels in the area outside the connected fusion region and in the area not contaminated by ink; calculate the gray median of the multiple background reference pixels as the background primary color value; and replace the pixel value of the pixel to be processed with the background primary color value to generate the independent marker image.

[0084] Furthermore, the offset calculation module 14 in the adaptive control system of the high-speed rotary printing press based on dynamic monitoring is also used for: extracting edge gradient distribution features based on the independent marker image; locating the continuous centroid coordinates of the alignment marks corresponding to the color plates by continuously fitting and deducing the edge gradient distribution features; selecting the alignment mark with the highest energy value as the origin mark based on the pixel energy features of the alignment marks, and setting the continuous centroid coordinates of the origin mark as the origin of the reference coordinate system; and calculating the physical offset distance of the continuous centroid coordinates of the alignment marks corresponding to other color plates relative to the origin of the reference coordinate system in the vertical paper feeding direction and along the paper feeding direction, based on the origin of the reference coordinate system, as the overprinting offset.

[0085] Furthermore, the overprinting adjustment module 15 in the adaptive control system of the high-speed rotary printing press based on dynamic monitoring is also used for: calculating an initial compensation step size based on the overprinting offset, combined with real-time operating parameters of printing speed and paper tension, wherein the overprinting offset includes a vertical paper feed direction offset component and a paper feed direction offset component, the vertical paper feed direction offset component corresponds to the calculated mechanical transmission compensation step size, and the paper feed direction offset component corresponds to the calculated control board signal compensation step size; generating the control command based on the initial compensation step size; setting a command silence suppression period based on the execution response delay time after the control command is issued, wherein the execution response delay time is determined comprehensively based on the mechanical transmission delay time corresponding to the mechanical transmission compensation step size and the electrical signal delay time corresponding to the control board signal compensation step size, blocking the generation of the next round of control commands during the command silence suppression period until the overprinting offset is resampled and verified to converge to a preset convergence threshold.

[0086] Furthermore, the overprinting adjustment module 15 in the adaptive control system of the high-speed rotary printing press based on dynamic monitoring is also used to: continuously acquire a preset number of re-sampling images after the instruction silence period expires, calculate the real-time overprinting offset corresponding to the re-sampling images respectively; calculate the temporal variance of the real-time overprinting offset; when the temporal variance is less than a preset stability threshold and the mean of the real-time overprinting offset is within a preset convergence threshold range, determine convergence and release the blockade on the generation of the next round of control instructions.

[0087] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The adaptive control method and specific examples of the high-speed rotary printing press based on dynamic monitoring in the aforementioned embodiment 1 are also applicable to the adaptive control system of the high-speed rotary printing press based on dynamic monitoring in this embodiment. Through the foregoing detailed description of the adaptive control method of the high-speed rotary printing press based on dynamic monitoring, those skilled in the art can clearly understand the adaptive control system of the high-speed rotary printing press based on dynamic monitoring in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0088] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0089] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. An adaptive control method for high-speed rotary printing presses based on dynamic monitoring, characterized in that, include: Acquire a sheet image during the high-speed printing process, the sheet image containing reference line features and alignment marks; Based on the image, the connected and fused region formed by the dynamic diffusion of ink between the reference line features and the alignment mark is identified; Based on the connected fusion region, the alignment markers are reconstructed using anti-diffusion features to obtain independent marker images; Based on the independent marked image, calculate the overprinting offset between each color plate; Based on the overprinting offset, a control command is generated and sent to the printhead execution module to perform adaptive overprinting adjustment of the printing press; Based on the image, the connected and merged region formed by the dynamic diffusion of ink between the reference line features and the alignment mark is identified, including: Based on the image, a linear region extending along the printing paper feed direction and having a continuous high gray value is extracted as the main feature region of the reference line. Connectivity growth detection is performed outward from the edge of the main feature region of the reference line; The region where the growth detection shows a spatial intersection with the alignment mark and a continuous gray-level gradient without any breaks is determined to be the connected and fused region. Based on the independently marked image, the overprinting offset between each color plate is calculated, including: Based on the independently labeled image, extract edge gradient distribution features; By continuously fitting and deducing the edge gradient distribution characteristics, the continuous centroid coordinates of the corresponding alignment marks of the excellent version are located respectively. Based on the pixel energy characteristics of the alignment mark, the alignment mark with the highest energy value is selected as the origin mark, and the continuous centroid coordinates of the origin mark are set as the origin of the reference coordinate system. Based on the origin of the reference coordinate system, the physical offset distance of the continuous centroid coordinates of the alignment marks corresponding to other color plates relative to the origin of the reference coordinate system in the direction perpendicular to the paper feed and along the paper feed direction is calculated, and this offset is used as the overprinting offset.

2. The adaptive control method for a high-speed rotary printing press based on dynamic monitoring as described in claim 1, characterized in that, Before obtaining an independent marker image by reconstructing anti-diffusion features of the alignment markers based on the connected fusion region, the process includes: Based on the connected fusion region, a one-dimensional grayscale profile curve is extracted along the normal direction pointing from the reference line feature to the alignment mark. The one-dimensional grayscale profile curve is differentiated to obtain the grayscale gradient change rate curve of ink diffusion energy attenuation. Based on the gray-level gradient change rate curve, the local extreme points of the gray-level gradient change rate are captured as the physical boundaries of the alignment marker.

3. The adaptive control method for a high-speed rotary printing press based on dynamic monitoring as described in claim 2, characterized in that, Based on the connected fusion region, the alignment markers are reconstructed using anti-diffusion features to obtain independent marker images, including: Pixels located outside the physical boundary and within the connected fusion region are designated as pixels to be processed. The pixel values ​​of the pixels to be processed are replaced with the background base color values ​​to generate the reconstructed independent marker image.

4. The adaptive control method for a high-speed rotary printing press based on dynamic monitoring as described in claim 3, characterized in that, The pixel values ​​of the pixels to be processed are replaced with the background primary color values ​​to generate the reconstructed independent marker image, including: In the area outside the connected and fused region and uncontaminated by ink, multiple background reference pixels are extracted; Calculate the median grayscale value of the plurality of background reference pixels, and use it as the background primary color value; The pixel values ​​of the pixels to be processed are replaced with the background base color values ​​to generate the independent marker image.

5. The adaptive control method for a high-speed rotary printing press based on dynamic monitoring as described in claim 1, characterized in that, Based on the aforementioned overprinting offset, a control command is generated and sent to the printhead execution module to perform adaptive overprinting adjustment of the printing press, including: Based on the registration offset, the initial compensation step size is calculated by combining the real-time operating parameters of printing speed and paper tension. The registration offset includes a vertical paper feed direction offset component and a paper feed direction offset component. The vertical paper feed direction offset component corresponds to the calculation of the mechanical transmission compensation step size, and the paper feed direction offset component corresponds to the calculation of the control board signal compensation step size. The control command is generated based on the initial compensation step size; Based on the execution response delay time after the control command is issued, a command silence suppression period is set. The execution response delay time is determined by combining the mechanical transmission delay time corresponding to the mechanical transmission compensation step size and the electrical signal delay time corresponding to the control board signal compensation step size. During the command silence suppression period, the generation of the next round of control commands is blocked until the overprinting offset is verified to converge to the preset convergence threshold.

6. The adaptive control method for a high-speed rotary printing press based on dynamic monitoring as described in claim 5, characterized in that, During the instruction silence period, the generation of the next round of control instructions is blocked until the overlay offset is verified to converge to a preset convergence threshold, including: After the instruction silence period expires, a preset number of re-acquired images are continuously acquired, and the real-time overprint offset corresponding to the re-acquired images is calculated respectively. Calculate the temporal variance of the real-time overprint offset; When the time series variance is less than a preset stability threshold and the mean of the real-time overprint offset is within a preset convergence threshold range, convergence is determined, and the blockade on the generation of the next round of control instructions is lifted.

7. The adaptive control method for a high-speed rotary printing press based on dynamic monitoring as described in claim 1, characterized in that, Acquiring images of the printing format during high-speed printing, including: Acquire the initial image of the high-speed rotary printing press in motion and the corresponding real-time printing speed data; Based on the real-time printing speed data, a motion blur degradation matrix is ​​constructed by matching a preset dynamic point expansion function model. The initial image is restored by Wiener filtering using the motion blur degradation matrix to generate the image with motion blur eliminated.

8. A high-speed rotary printing press adaptive control system based on dynamic monitoring, characterized in that, The step of implementing the adaptive control method for a high-speed rotary printing press based on dynamic monitoring according to any one of claims 1 to 7, wherein the adaptive control system for a high-speed rotary printing press based on dynamic monitoring comprises: A format image acquisition module is used to acquire format images during high-speed printing, wherein the format images include reference line features and alignment marks; The connected and blended region identification module is used to identify, based on the image, the connected and blended region formed by the dynamic diffusion of ink between the reference line features and the alignment mark; The anti-diffusion feature reconstruction module is used to reconstruct the anti-diffusion features of the alignment markers based on the connected fusion region to obtain an independent marker image; The offset calculation module is used to calculate the overprint offset between each color plate based on the independent marked image; The overprinting adjustment module is used to generate and send control commands to the printhead execution module based on the overprinting offset to perform adaptive overprinting adjustment of the printing press.

Images