A CTP plate making quality detection and analysis method based on image recognition

Abstract

The application discloses a kind of CTP plate making quality detection analysis method based on image recognition technical field, it is related to the field of image recognition CTP plate making quality detection, including: using high-resolution image acquisition equipment Synchronous acquisition has been made CTP plate image and the printed matter image produced by the plate;With the geometric center outside the effective printing area of plate and corner point as physical reference mark, a unified coordinate system is constructed, and feature positioning and area division are carried out based on the unified coordinate system;By extracting multi-dimensional feature data such as dot gain and dirty dot coverage, generate deviation data set, and combine feature type and position relationship to carry out combined logic analysis to determine the reason of dot overexpansion and the root cause of dirty dot generation, realize the automatic detection and root cause tracing of CTP plate making quality, improve the detection efficiency.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of CTP plate making quality based on image recognition, and relates to a CTP plate making quality detection and analysis method based on image recognition. BACKGROUND

[0002] CTP stands for computer-to-plate technology, which is a modern plate making process that directly outputs digital graphic information to printing plates through laser or other imaging methods. It has been widely used in the printing industry. In the CTP plate making process, the quality indicators such as the dot precision and the contamination of the printing plate directly determine the clarity, color consistency and neatness of the final printed product. Once the printing plate has quality defects, it will lead to the rejection of batch printed products and cause serious economic losses. Therefore, quality monitoring and analysis based on image recognition has become a core technology in the CTP plate making process.

[0003] At present, there are some existing technologies for CTP plate making quality detection based on image recognition. For example, the invention patent with the publication number CN101556251A proposes a CTP plate making quality detection method based on a digital signal processor. The method uses a specific structure to collect images of the dots, lines and chessboard regions on the CTP plate control strip, adjusts the relevant components to obtain microscopic imaging at a fixed object distance, extracts the dot features of the ladder region, the straight line features of the straight line region and the chessboard block features of the chessboard region from the central sub-image, and performs comprehensive evaluation based on the membership functions of the features combined with the weights. The results are displayed on the LCD screen, realizing objective detection of CTP plate making quality, replacing traditional manual subjective evaluation and improving detection efficiency and consistency.

[0004] Although the above-mentioned existing scheme has achieved certain results in the CTP plate making quality detection based on image recognition, it still has the following shortcomings: first, the image on the printed product comes entirely from the printing plate. Any abnormal point on the printed product is caused by the printing plate or the printing process. The existing technology only detects the image of the printing plate, which is a isolated quality control link, and cannot trace the quality through the mapping relationship between the printing plate and the printed product. Therefore, it is difficult to accurately determine whether the quality problem in the printing process is caused by the plate making process or the printing process, reducing the comprehensiveness and accuracy of quality analysis.

[0005] Secondly, the existing technology only adjusts the object distance to obtain a fixed object distance for image collection, and performs statistical measurement based on a small sub-image. The analysis of the features of each region is independent of each other, and lacks positional correlation. Therefore, after detecting the feature abnormalities, it is difficult to accurately locate the specific position of the abnormalities on the printing plate, which brings great difficulties to the targeted repair of the printing plate and the elimination of unqualified areas, reducing the utilization rate of the printing plate.

[0006] In addition, existing technologies can only provide a binary judgment of pass or fail by comparing the extracted halftone dots, lines, and checkerboard features with membership functions, generating a defect report pointing to a specific process step. This means that after a quality anomaly is detected, multiple possible influencing factors such as exposure parameters, plate quality, equipment precision, and operating procedures must be checked one by one, making it impossible to quickly pinpoint the root cause of the problem and reducing overall production efficiency. Summary of the Invention

[0007] In view of this, in order to solve the problems mentioned in the background art, the present invention provides a CTP plate-making quality detection and analysis method based on image recognition.

[0008] The objective of this invention can be achieved through the following technical solution: a CTP plate-making quality detection and analysis method based on image recognition, comprising: S1, using a high-resolution image acquisition device to simultaneously acquire images of the prepared CTP printing plate and images of the printed matter produced by the printing plate.

[0009] S2. After preprocessing the acquired images of printing plates and printed matter, a unified coordinate system for the images of printing plates and printed matter is established based on the physical correspondence between the printing plates and printed matter.

[0010] S3. Identify and segment individual halftone dot regions and dirt regions on the two images respectively, and extract halftone geometric features and dirt shape features to form a printing plate feature dataset and a printed matter feature dataset.

[0011] S4. Based on a unified coordinate system, map and match the coordinates of the geometric features of the halftone dots in the printed matter feature dataset with the printing plate feature dataset, calculate the halftone dot gain rate for each corresponding halftone dot, and statistically analyze the dirt coverage rate of the corresponding area to generate a deviation dataset containing location information.

[0012] S5. Generate a quality defect root cause quality report based on the deviation dataset and the combination logic of feature position relationship and feature type.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention uses a high-resolution image acquisition device to simultaneously acquire the image of the CTP printing plate that has been made and the image of the printed matter produced by the printing plate. It can trace the source of quality through the correlation analysis of the printing plate and the printed matter, accurately determine whether the quality problem originates from the plate-making stage or the printing stage, and effectively solve the problem of insufficient comprehensiveness and accuracy of quality analysis.

[0014] (2) The present invention constructs a unified coordinate system by using the geometric center and corner points outside the effective printing area of ​​the printing plate as physical reference marks, and performs feature positioning and area division based on the unified coordinate system, so that the feature analysis of each area has clear positional correlation, which greatly facilitates the targeted repair of the printing plate and the removal of unqualified areas, and improves the utilization rate and detection efficiency of the printing plate.

[0015] (3) The application generates a deviation data set by extracting multi-dimensional feature data such as dot expansion rate and dirty dot coverage, and combines feature types and position relationships to perform combined logic analysis to determine the reasons for dot overexpansion and the root causes of dirty dots, without checking multiple influencing factors one by one, thereby realizing rapid locking of quality defect roots. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0017] Figure 1 The method embodiment of the application is shown in the flowchart.

[0018] Figure 2 The process chart for establishing a unified coordinate system of the CTP plate and printed matter of the application is shown.

[0019] Figure 3 The flowchart for dirty dot detection and feature extraction of the binary image of the application is shown. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only some embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the application.

[0021] Please refer to Figure 1 As shown in the drawings, the application provides a CTP plate making quality detection and analysis method based on image recognition, which comprises the following steps: S1, using a high-resolution image acquisition device to synchronously acquire an image of a CTP plate that has been made and an image of printed matter produced by the plate.

[0022] It should be noted that the high-resolution image acquisition device can be an industrial area array camera, which is matched with a white LED flat light source to adjust the angle and intensity of the light source, so as to ensure that the acquired image has no reflection and no shadow.

[0023] The prepared CTP printing plate is fixed on a high-precision stage, and it is ensured that the printing plate is flat and wrinkle-free, and a printed product produced by the printing plate under standard printing process parameters is also fixed on the same stage at an adjacent position. The camera is controlled by a trigger controller to synchronously capture the printing plate and the printed product, and high-resolution images of the two are obtained. Synchronous acquisition can avoid image distortion caused by time difference or environmental changes, and improve the accuracy of comparison.

[0024] S2, after preprocessing the collected printing plate and printed product images, a unified coordinate system of the printing plate image and the printed product image is established based on the physical correspondence between the printing plate and the printed product.

[0025] It should be noted that image preprocessing includes but is not limited to denoising, grayscale, contrast enhancement, etc., so as to improve image quality, unify processing standards, and provide reliable basic data for subsequent core detection steps.

[0026] Referring to Figure 2 As shown in the figure, one specific implementation process of establishing a unified coordinate system of the printing plate image and the printed product image is as follows: geometric center and corner points outside the effective printing area of the printing plate are marked as physical reference marks during plate making.

[0027] In the preprocessed printing plate image and printed product image, a plurality of non-collinear physical reference marks are identified.

[0028] It should be noted that using a plurality of reference marks effectively reduces the influence of single mark identification error on overall conversion accuracy, making the coordinate mapping of the printing plate and the printed product more stable and reliable.

[0029] The pixel coordinates of each physical reference mark in the printing plate image and the corresponding pixel coordinates in the printed product image are obtained.

[0030] Based on the two groups of pixel coordinates of the physical reference marks in the printing plate image and the printed product image, a coordinate conversion matrix of the printed product image coordinates to the printing plate image coordinates is constructed, and the pixel coordinates of all feature points in the printed product image are uniformly converted to the coordinate system of the printing plate image.

[0031] It should be noted that the specific implementation steps of constructing the coordinate conversion matrix are as follows: the pixel coordinates of the reference marks in the printing plate image and the pixel coordinates of the corresponding reference marks in the printed product image are recorded.

[0032] An affine transformation model is used to solve the conversion matrix between the two groups of coordinates by least squares method, which contains rotation, translation and scaling parameters.

[0033] The vertex of the printing plate image is selected as the origin of the unified coordinate system.

[0034] Specifically, the coordinates of all pixels in the printed image are converted into the coordinates of the printing plate image by point multiplication operation with the conversion matrix, and the top-left vertex of the printing plate image is taken as the origin of the unified coordinate system.

[0035] It should be noted that converting all feature points in the printed image into the coordinate system of the printing plate image enables subsequent feature matching, deviation analysis and other operations to be performed in the same space, facilitates unified management of position information, and avoids matching errors caused by inconsistent coordinate systems.

[0036] S3, respectively, identify and segment single dot areas and dirty dot areas on the two images, and extract dot geometric features and dirty dot shape features to form a printing plate feature data set and a printed product feature data set.

[0037] A specific implementation step of segmenting single dot areas and dirty dot areas is as follows: perform connected region detection on the preprocessed printing plate image and the printed image, and calculate the average gray value of each connected region pixel.

[0038] The specific steps of the connected region detection are as follows: first, traverse all pixels of the preprocessed printing plate image and the printed image row by row and column by column, judge the adjacent pixels of each unmarked pixel, and usually use 8-neighborhood rule, that is, whether the pixels in the upper, lower, left, right and four diagonal directions satisfy the connectivity condition with the pixel, that is, the gray value difference of adjacent pixels is within a preset range.

[0039] Next, if a pixel and its adjacent pixels satisfy the connectivity condition, these pixels are classified into the same connected region, and a unique identification number is assigned to the region. The marked pixels are not detected repeatedly, and the whole image connected region is identified until all pixels are traversed.

[0040] Finally, check whether the same region is split due to the pixel traversal order for the preliminarily identified connected region, and merge the adjacent small regions that meet the merging conditions.

[0041] If the average gray value falls within the preset dot gray value interval, the connected region is marked as a dot region and segmented into an independent dot unit.

[0042] It should be noted that the preset dot gray value interval is determined by statistical analysis of a large number of known samples to determine the central distribution range of the dot gray value.

[0043] The gray value of a pixel in a gray image represents the brightness of the point. In printing, the dot is the area where ink is attached, and its reflectivity is low, so it appears darker in the gray image. The blank substrate of the printing plate has high reflectivity and appears brighter in the image. Converting the image segmentation problem into a data-based statistical judgment problem avoids subjective speculation.

[0044] The pre-processed printing plate image and the printed image are converted into gray images and then binarized.

[0045] The area with a gray value greater than the preset gray threshold value after binarization is divided into a foreground pixel area, and vice versa.

[0046] It should be noted that the foreground pixel area, i.e. the dirty spot pixel area, is usually oil stains, dust, scratches, ink skin, etc. They physically block light reflection or absorb more light, so they appear dark on the image, i.e. with a lower gray value. The background pixel area is the normal printing plate substrate or printed paper, which has a smooth surface and high reflectivity, so it appears bright on the image, i.e. with a higher gray value. The inherent gray difference between the dirty spot and the background is used to distinguish them.

[0047] The preset gray threshold value traverses all possible threshold values, calculates the intra-class variance of the foreground and background pixels segmented based on the threshold value, and selects the threshold value that maximizes the inter-class variance as the best threshold value.

[0048] If the area of the foreground pixel area exceeds the noise threshold, it is marked as a dirty spot area, and the confirmed dirty spot area is segmented to generate independent dirty spot units.

[0049] It should be noted that the dirty spot usually has a certain physical size, which is reflected in the image as a continuous area composed of many pixels. Their area, i.e. the number of pixels, will exceed a minimum threshold value that can be considered a defect.

[0050] The noise threshold is determined based on the theoretical resolution of the imaging system and the statistical analysis of the noise of a large number of good images.

[0051] The dot geometric feature extraction step is as follows: the connected region with a gray mean value falling within a preset dot gray value interval is marked as a target connected domain.

[0052] The total number of pixels contained in the target connected domain is counted to determine the area parameter of the dot.

[0053] The geometric center of the connected domain is calculated through the contour boundary pixel coordinates, and the position coordinates of the center in the unified coordinate system are recorded.

[0054] Specifically, let the connected domain be a finite two-dimensional discrete point set, and the area of the connected domain In the formula, R represents the set of pixel coordinates in the connected region, N represents the cardinality of the set, and i represents the i-th pixel. .

[0055] Sum of x-coordinate components In the formula, This represents the column coordinate of the i-th pixel, and the sum of its y-coordinate components. In the formula, This represents the row coordinate of the i-th pixel.

[0056] Geometric center x-coordinate ordinate of geometric center .

[0057] The area parameters and center coordinates of the dots are defined as the geometric features of the dots.

[0058] In the printing industry, dot gain is one of the most crucial indicators for measuring print quality, directly manifested as an increase in dot area. Using area as the primary characteristic directly reflects the core issue in quality inspection.

[0059] The center coordinates provide the absolute position information of each dot, which is the basis for subsequent mapping and matching between the printing plate and the printed image.

[0060] See Figure 3 As shown, the steps for extracting the shape features of the dirt spot are as follows: scan the binary image to identify the continuous region formed by all foreground pixels, and regard each continuous region as a candidate dirt spot.

[0061] For each candidate dirty point, perform contour tracking to obtain its boundary contour pixel set.

[0062] Specifically, the entire image is traversed from left to right and from top to bottom to find the first pixel with a value of 1. This pixel must be one of the top left boundary points of a connected region, and it is used as the starting point for contour tracking. An initial movement direction is set, usually assumed to be from the left background to the foreground point, so the initial search direction is upward.

[0063] Starting from the current boundary point and the current movement direction, examine the pixels in its 8-neighborhood in a clockwise or counterclockwise order, centering on the current point, starting from the direction preceding the current movement direction. For example, if the previous movement was to the right, then this time the examination would start from the upper right and proceed clockwise. Once a foreground pixel is found in the neighborhood, that pixel becomes the next boundary point. Record the coordinates of this new point in the contour set, update the current point, and update the current movement direction.

[0064] Repeat the above steps until both of the following conditions are met simultaneously, then stop tracking: the next boundary point returns to the starting point, and the next moving direction is the same as the initial moving direction.

[0065] At this time, all recorded point coordinates constitute an ordered, closed contour pixel set.

[0066] Traverse the contour pixel set to determine its extreme points distributed on a two-dimensional plane, and fit a minimum area rectangle that can completely surround all contour pixels according to the positions of the extreme points.

[0067] Count the total number of pixels in each candidate dirty point region as the area of the dirty point.

[0068] Calculate the geometric center point of each candidate dirty point, and record the position coordinates of the center point in the unified coordinate system.

[0069] Define the area and center coordinates of the dirty point as the shape features of the dirty point.

[0070] S4, based on the unified coordinate system, map and match the coordinates of the dot geometric features in the printing product feature data set with the printing plate feature data set, calculate the dot expansion rate of each corresponding dot one by one, and count the dirty point coverage rate of the corresponding area to generate a deviation data set containing position information.

[0071] The specific content of mapping and matching the coordinates of the dot geometric features in the printing product feature data set with the printing plate feature data set based on the unified coordinate system is as follows: select the dot or dirty point feature unit in sequence from the printing product feature data set, and read its position coordinates.

[0072] Based on the established coordinate conversion relationship, the position coordinates of the above printing product features are reversely mapped back to the coordinate system of the original printing plate image to obtain the expected printing plate coordinate position.

[0073] Taking the expected printing plate coordinate position as the center, search for all feature units within the preset geometric tolerance range in the printing plate feature data set.

[0074] It should be noted that the preset geometric tolerance range is mainly determined by statistical analysis of a large number of known correct matching pairs to determine a tolerance value that covers more than 99% of the deviations, or another way is to estimate the total error of the entire system chain.

[0075] If no printing plate feature unit is searched within the tolerance range, it is determined that the printing product feature has no corresponding source on the printing plate, and is recorded as an abnormal state of printing plate missing.

[0076] If multiple printing plate feature units are searched within the tolerance range, it is determined that the types of these units are consistent with the type of the current printing product feature.

[0077] If the types are consistent, the closest location of the plate feature unit is selected to determine the corresponding feature of the matched success, and a mapping relationship is established.

[0078] If the types are inconsistent, it is determined that the printed matter feature has no corresponding root of the same type on the plate, and is recorded as an abnormal state of type inconsistency.

[0079] Through the mapping matching in the unified coordinate system, the association channel between the printed matter dots and the plate dots is opened, which is not only the basis for calculating the dot expansion rate, but also the core link for identifying dot abnormalities and determining defect roots.

[0080] The dot expansion rate calculation step is as follows: based on the mapping matching result, a pair of dot feature units with established corresponding relationship in the plate feature data set and the printed matter feature data set is obtained.

[0081] For each matched dot unit, the area parameter of the dot is extracted from the plate feature data set, and the area parameter of the corresponding dot is extracted from the printed matter feature data set.

[0082] The area difference value is obtained by subtracting the area parameter of the printed matter dot from the area parameter of the corresponding plate dot, and the dot expansion rate value is obtained by dividing the area difference value by the area parameter of the plate dot.

[0083] The dot expansion rate value is associated with the corresponding dot position information and dot identification information.

[0084] The standard deviation of the expansion rate value of the dots in each region is calculated to generate statistical indicators describing the dot expansion of each region.

[0085] If the dot expansion rate is too small, it may cause the color of the printed matter to be too light and the tone to be lost, i.e. the dot in the dark tone area is not full. If the dot expansion rate is too large, it will cause the color to be too deep and the text to be blurred, i.e. the dot in the light tone area is merged. Only when the dot expansion rate is controlled within a reasonable range, the consistency of the printed matter and the digital original can be ensured. Therefore, calculating the dot expansion rate is the key to converting quality from subjective observation to objective data, and provides a standardized basis for quality determination.

[0086] The dirty dot coverage rate calculation step is as follows: according to the physical layout features of the plate and the printed matter, the effective area is divided into a plurality of independent regions with clear boundaries.

[0087] It should be noted that the physical layout features of the plate and the printed matter refer to the physical size and shape of the effective printing area.

[0088] The effective printing area of the plate in the unified coordinate system is taken as the division object, the boundary of the effective printing area of the plate is taken as the division range, and the rectangular grid according to the industry convention is taken as the standard unit of the independent region.

[0089] Obtain the feature data of all the dirty spots in each independent area and the mapping relationship state of the dirty spots on the printing plate, and distinguish the effective dirty spots with corresponding sources on the printing plate from the abnormal dirty spots without corresponding sources.

[0090] It should be noted that if there is dirt outside or at the edge of the effective area of the printing plate, it will be transferred to the corresponding position of the printed matter synchronously with the ink during the printing process, forming a dirty spot with a source on the printing plate. If there is no dirt on the printing plate, the dirty spot appearing on the printed matter can only be an exogenous pollution introduced during the printing process, such as paper dust, printing machine roller impurities, ink impurities, etc. Such dirty spots have no corresponding source on the printing plate.

[0091] The dirty spots with corresponding sources on the printing plate are identified as effective dirty spots and included in the coverage rate statistics, and the dirty spots without corresponding sources on the printing plate are marked as abnormal dirty spots and recorded separately.

[0092] It should be noted that the effective dirty spot is derived from the dirt on the printing plate, which is a direct manifestation of the quality defect in the CTP plate making link, and its coverage rate directly reflects the cleanliness level of the printing plate, which is a core index of plate making quality control and must be included in the coverage rate statistics for quantitative evaluation of printing plate problems.

[0093] The abnormal dirty spot is derived from exogenous pollution in the printing process, such as paper dust and printing machine impurities, which is irrelevant to the quality of the printing plate. If it is included in the coverage rate statistics, it will lead to distorted statistical results and interfere with the accurate evaluation of plate making quality.

[0094] The number of effective dirty spots in each independent area is counted, and the area parameter of the area is obtained to derive the dirty spot coverage rate of the area.

[0095] According to the dirty spot coverage rate of each area, the pollution level of the area is determined.

[0096] Specifically, if the dirty spot coverage rate of an area is less than a preset low threshold, such as 0.1%, it is determined to be slightly contaminated, if the coverage rate is between the preset low threshold and the preset high threshold, and the preset low threshold can be 0.1% and the preset high threshold can be 1%, it is determined to be moderately contaminated, and if the coverage rate is greater than the preset high threshold, such as 1%, it is determined to be severely contaminated.

[0097] The dirty spot coverage rate and pollution level of each area are output to form a complete area dirty spot coverage result.

[0098] The dirty spot coverage rate is converted from qualitative description to quantitative data, providing a standardized and reproducible basis for determining the pollution level, avoiding subjective judgment errors, and ensuring uniform evaluation standards for dirty spot pollution in different areas and different batches.

[0099] The bias data set contains the following specific content: the enlargement rate value and corresponding position coordinates of each dot obtained by dot mapping matching and enlargement rate calculation.

[0100] The area dot coverage and area position identification obtained by the dot mapping matching and coverage statistics.

[0101] The formation of the deviation data set containing position information is the only data source for defect root cause determination. By analyzing the numerical characteristics and spatial distribution of the data, it can accurately distinguish whether the defect root cause is a plate problem or a printing process problem, and finally provide structured and traceable data support for quality report generation.

[0102] S5, generating a quality defect root cause quality report according to the combination logic of the deviation data set, the feature position relationship and the feature type.

[0103] The specific content of the quality defect root cause quality report is as follows: extracting each dot expansion rate and corresponding position, each area dot coverage and area position information from the deviation data set, and dividing into dot deviation data and dot deviation data according to the feature type.

[0104] If most of the dot expansion rates in a certain area are similar and higher than those in other areas, it is determined that the printing pressure or ink amount is abnormal, causing the dot to expand excessively.

[0105] It should be noted that if the printing pressure of a certain area is too large, the contact pressure between the plate and the rubber blanket, and the rubber blanket and the printing material will increase as a whole, resulting in an increase in ink transfer amount of all dots in the area, and thus the dot area expands generally, and the expansion degree tends to be similar due to the uniformity of the pressure.

[0106] If the ink supply amount of a certain area is too much, the ink will spread excessively at the edge of the dot, causing all dots in the area to expand at the edge, and because the ink supply system has consistency in ink supply amount to the same area, the dot expansion rate will show the characteristics of overall high and close in value.

[0107] If the difference between the expansion rate of a single dot and the average expansion rate of the surrounding dots exceeds the set deviation threshold, and the expansion rate of the surrounding dots is normal, it is determined that the corresponding plate area is worn or ink is accumulated.

[0108] It should be noted that the set deviation threshold is usually determined by statistical process control method.

[0109] The difference from the average expansion rate of the surrounding dots exceeds the set deviation threshold, indicating that the behavior of the dot deviates significantly from the normal trend of the region it belongs to. In statistics, this is an outlier.

[0110] Systematic factors such as printing pressure and ink flowability that cause dot expansion usually affect an area. If the expansion rate of a dot is much higher than that of its adjacent dots, it means that the systematic pressure or ink amount setting is not the main reason for its abnormality. Otherwise, the surrounding dots should also show similar abnormalities.

[0111] If the coverage of the dirty spot in the area of the printed matter is high and there is a similar dirty spot in the corresponding area of the printing plate, it is determined that the dirty spot in the printed matter is caused by the dirty printing plate.

[0112] It should be noted that if there is a dirty spot in a certain area of the printing plate, the dirty spot in the area will be adsorbed with ink under the action of printing pressure and will be transferred to the same spatial position of the printed matter, forming a dirty spot in the printed matter consistent with the characteristics of the dirty printing plate. On the contrary, if there is no dirty spot on the printing plate, the dirty spot in the area of the printed matter can only be caused by external sources such as paper dust and equipment impurities during the printing process, rather than by the printing plate.

[0113] If the dirty spot in the printed matter has no corresponding printing plate and is concentrated on the edge of the image, it is determined that the printing process is contaminated by external impurities.

[0114] It should be noted that if there is no corresponding printing plate, it means that there is no similar feature on the printing plate, which directly excludes the source of the dirty printing plate transfer and proves that it is a newly added non-printing plate source defect during the printing process.

[0115] The parameters involved in the above formula are de-dimensioned to calculate their numerical values. The formula is obtained by software simulation of a large amount of data to reflect the most real situation. The preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0116] The above embodiments can be realized wholly or partially by software, hardware, firmware or any combination thereof. When realized by software, the above embodiments can be realized wholly or partially in the form of a computer program product.

[0117] Those skilled in the art can realize that the modules and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized by hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0118] In addition, the functional modules in each embodiment of the present application can be integrated in one processing module, or each module can exist physically, or two or more modules can be integrated in one module.

[0119] The above is merely a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0120] Finally, the above only the preferred embodiments of the present application, and not for limiting the present application, any modifications, made within the spirit and principles of the present application, equivalent replacement, improvement, etc., should be included within the scope of the present application.

Claims

1. An image recognition based CTP plate making quality detection and analysis method, characterized in that: The method comprises the following steps: S1, using a high-resolution image acquisition device to synchronously acquire images of a manufactured CTP printing plate and printed matter produced by the printing plate; S2, after preprocessing the acquired printing plate and printed matter images, establishing a unified coordinate system of the printing plate image and the printed matter image based on the physical correspondence between the printing plate and the printed matter; S3, identifying and segmenting single dot areas and dirty dot areas on the two images respectively, and extracting dot geometric features and dirty dot shape features to form a printing plate feature data set and a printed matter feature data set; S4, mapping and matching the coordinates of the dot geometric features in the printed matter feature data set with the printing plate feature data set based on the unified coordinate system, calculating the dot expansion rate of each corresponding dot one by one, and counting the dirty dot coverage rate of the dirty dot area to generate a deviation data set containing position information; S5, generating a quality defect root cause quality report according to the deviation data set combined with the combination logic of feature position relationship and feature type.

2. The CTP plate making quality detection and analysis method based on image recognition according to claim 1, characterized in that: The specific steps of establishing the unified coordinate system of the printing plate image and the printed matter image are as follows: During plate making, geometric centers and corner points outside the effective printing area of the printing plate are marked as physical reference marks; In the preprocessed printing plate image and printed matter image, a plurality of non-collinear physical reference marks are identified respectively; The pixel coordinates of each physical reference mark in the printing plate image and the corresponding pixel coordinates in the printed matter image are obtained; Based on the two groups of pixel coordinates of the physical reference marks in the printing plate image and the printed matter image, a coordinate conversion matrix of the printed matter image coordinates to the printing plate image coordinates is constructed, and the pixel coordinates of all feature points in the printed matter image are uniformly converted to the coordinate system of the printing plate image; The vertex of the printing plate image is selected as the origin of the unified coordinate system.

3. The CTP plate quality detection and analysis method based on image recognition according to claim 1, characterized in that: The specific steps of segmenting single dot areas and dirty dot areas are as follows: Connected region detection is performed on the preprocessed printing plate image and printed matter image, and the pixel gray mean value of each connected region is calculated; If the gray mean value falls within the preset dot gray value interval, the connected region is marked as a dot region and segmented into independent dot units; After the preprocessed printing plate image and printed matter image are converted into gray scale images, binary processing is performed; Regions with gray values greater than the preset gray threshold value after binary processing are divided into background pixel regions, and vice versa for foreground pixel regions; If the area of the foreground pixel region exceeds the noise threshold, it is marked as a dirty dot region, and the independent dirty dot unit is generated by segmenting the confirmed dirty dot region.

4. The CTP plate quality detection and analysis method based on image recognition according to claim 3, characterized in that: The dot geometric feature extraction steps are as follows: The connected region with a gray mean value falling within the preset dot gray value interval is marked as a target connected domain; The total number of pixels contained in the target connected domain is counted to determine the area parameter of the dot; The geometric center of the connected domain is calculated through the contour boundary pixel coordinates, and the position coordinates of the center in the unified coordinate system are recorded; The area parameter and center coordinates of the dot are defined as dot geometric features.

5. The CTP plate quality detection and analysis method based on image recognition according to claim 1, characterized in that: The dirty dot shape feature extraction steps are as follows: Scan the binary image to identify all continuous regions formed by foreground pixels, and consider each continuous region as a candidate dirty dot; Contour tracking is performed on each candidate dirty dot to obtain its boundary contour pixel set; The extreme value points of the distribution of the contour pixel set on the two-dimensional plane are determined by traversing the contour pixel set, and a minimum area rectangle completely surrounding all the contour pixels is fitted according to the positions of the extreme value points; The total number of pixels in each candidate dirty point region is counted as the area of the dirty point; The geometric center point of each candidate dirty point is calculated, and the position coordinates of the center point in the unified coordinate system are recorded; The area and center coordinates of the dirty point are defined as the shape characteristics of the dirty point.

6. The CTP plate quality detection and analysis method based on image recognition according to claim 1, characterized in that: The mapping matching of the coordinates of the dot geometric characteristics in the printing product feature data set and the printing plate feature data set based on the unified coordinate system is specifically as follows: The dot or dirty point feature units are sequentially selected from the printing product feature data set, and the position coordinates thereof are read; The position coordinates of the above printing product features are reversely mapped back to the coordinate system of the original printing plate image based on the established coordinate conversion relationship, to obtain the expected printing plate coordinate position; Taking the expected printing plate coordinate position as the center, all feature units within a preset geometric tolerance range in the printing plate feature data set are searched; If no printing plate feature unit is searched within the tolerance range, it is determined that the printing product feature has no corresponding root on the printing plate, and the abnormal state of printing plate missing is recorded; If multiple printing plate feature units are searched within the tolerance range, it is determined whether the types of these units are consistent with the type of the current printing product feature; If the types are consistent, the printing plate feature unit closest to the position is selected as the matched corresponding feature, and a mapping relationship is established; If the types are inconsistent, it is determined that the printing product feature has no corresponding root of the same type on the printing plate, and the abnormal state of type inconsistency is recorded.

7. The CTP plate quality detection and analysis method based on image recognition according to claim 4, characterized in that: The dot expansion rate calculation step is as follows: Based on the mapping matching result, a pair of dot feature units having established a corresponding relationship in the printing plate feature data set and the printing product feature data set is obtained; The area parameter of the dot in the printing plate feature data set is extracted for each matched dot unit, and the area parameter of the corresponding dot in the printing product feature data set is extracted; The area difference value is obtained by subtracting the area parameter of the printing product dot from the area parameter of the corresponding printing plate dot, and the dot expansion rate value is obtained by dividing the area difference value by the area parameter of the printing plate dot; The dot expansion rate value is associated with the corresponding dot position information and dot identification information; The standard deviation of the expansion rate values of the dots in each region is calculated to generate statistical indicators describing the dot expansion in each region.

8. The CTP plate quality detection and analysis method based on image recognition according to claim 1, characterized in that: The dirty point coverage rate calculation step is as follows: According to the physical layout characteristics of the printing plate and the printing product, the effective area is divided into a plurality of independent regions with clear boundaries; The feature data of all printing product dirty points and the mapping relationship state of the printing plate dirty points in each independent region are obtained, and the effective dirty points having corresponding roots on the printing plate and the abnormal dirty points having no corresponding roots are distinguished; The dirty points having corresponding roots on the printing plate are identified as effective dirty points and included in the coverage rate statistical range, and the dirty points having no corresponding roots on the printing plate are marked as abnormal dirty points and recorded separately; The number of effective dirty points in each independent region is counted to obtain the dirty point coverage rate of the region in combination with the area parameter of the region; The dirty point contamination level of each region is determined according to the dirty point coverage rate of the region; The dirty point coverage rate value and the contamination level of each region are output to form a complete regional dirty point coverage result.

9. The CTP plate quality detection and analysis method based on image recognition according to claim 1, characterized in that: The bias data set specifically contains the following: The expansion rate values and corresponding position coordinates of each screen dot obtained by screen dot mapping matching and expansion rate calculation; The coverage rate of each area dirty dot and the area position identifier obtained by dirty dot mapping matching and coverage rate statistics.

10. The CTP plate quality detection and analysis method based on image recognition according to claim 1, characterized in that: The quality defect root quality report specifically contains the following: Extract the screen dot expansion rate and corresponding position, the coverage rate of each area dirty dot and the area position information from the bias data set, and divide them into screen dot bias data and dirty dot bias data according to the feature type; If the expansion rates of most screen dots in a certain area are similar and higher than those in other areas, it is determined that the screen dots are excessively expanded due to abnormal printing pressure or ink amount; If the difference between the expansion rate of a single screen dot and the average expansion rate of the surrounding screen dots exceeds the set deviation threshold, and the expansion rates of the surrounding screen dots are normal, it is determined that the corresponding printing plate area is worn or ink is accumulated; If the coverage rate of a certain area dirty dot of the printed matter is high and the corresponding printing plate area has the same type of dirty dot, it is determined that the printing plate is dirty, causing the printed matter to be dirty; If the dirty dot of the printed matter has no corresponding printing plate and is concentrated on the image edge, it is determined that the printing process is contaminated by external impurities.

Images