A printing quality detection and analysis system based on machine vision

Abstract

The present invention belongs to the technical field of printed matter quality inspection, and specifically relates to a printed matter printing quality inspection and analysis system based on machine vision. By selecting samples of batch cigarette box prints and performing imaging tests based on the samples, suitable imaging parameters are determined according to the test results, and then the apparent images of the batch cigarette box prints are collected according to these parameters. This process can maximize the collection quality of the apparent image and provide reliable data support for apparent defect identification. At the same time, while the apparent images of the batch cigarette box prints are collected using suitable imaging parameters, infrared images are also collected simultaneously. By combining the apparent defects identified by the apparent images and the internal structural defects identified by the infrared images, the system can perform comprehensive printing quality evaluation and realize comprehensive defect identification. This method not only improves the accuracy and completeness of defect detection.

Description

Technical Field

[0001] The present invention belongs to the technical field of printed matter quality detection, and in particular relates to a printed matter printing quality detection and analysis system based on machine vision. Background Art

[0002] Despite the rise of digital media transforming how information is disseminated, printed materials still hold an irreplaceable place in modern society. This is particularly true in the tobacco packaging sector, where printed cigarette cases are in high demand. To ensure the quality and compliance of printed cigarette cases, rigorous quality inspections are essential after production. This process typically relies on high-precision image capture and analysis using visual inspection equipment to ensure compliance with stringent industry standards and quality requirements.

[0003] A prior art Chinese invention patent, published with publication number CN101738397A, discloses a printed product quality inspection process. This process uses visual inspection equipment to capture images of printed products. When defects are detected, the defect's location is recorded and the printed product is fed to a rewinding device for rewinding. The rewinding device compares the stored defect location information with the rewinding location information and stops rewinding when the printed product reaches the defect location. This invention separates detection and processing, reducing the burden on the visual inspection equipment and thus improving the efficiency of printed product quality inspection. However, this solution neglects adaptive adjustment of imaging parameters during the image acquisition process. Since printed products may have varying surface conditions due to factors such as raw materials and inks during production, without adaptive adjustment of imaging parameters, the captured images may exhibit overexposure, underexposure, blur, reflections, dark spots, and other issues, resulting in poor image quality and affecting accurate defect identification. For mass-produced printed products, this inaccurate image acquisition can lead to a large number of defects being missed or falsely detected, seriously impacting overall quality control of the printed products. In addition, low-quality images may increase the difficulty of subsequent data analysis and reduce the reliability and accuracy of the detection system.

[0004] Furthermore, the aforementioned solution, which uses visual inspection equipment to capture surface images of printed materials for defect identification, is limited to surface defects and cannot identify internal structural defects. Cigarette packaging is typically made of thick paper, making internal structural defects such as defect thickness and depth impossible to detect through surface images. Therefore, relying solely on surface image capture of printed materials cannot achieve comprehensive defect identification, significantly impacting the integrity and accuracy of quality inspection results. Summary of the Invention

[0005] The present invention aims to address the deficiencies in the prior art and provides a printed matter printing quality inspection and analysis system based on machine vision. By acquiring visible light and infrared images of mass-produced cigarette boxes, and selecting samples for each batch of cigarette box prints during the visible light image acquisition process for imaging tests, the optimal imaging parameters are determined. This achieves high-quality image acquisition and comprehensive defect identification for batch cigarette box prints, effectively making up for the problems existing in the prior art.

[0006] The purpose of the present invention can be achieved through the following technical solutions: A printed matter printing quality detection and analysis system based on machine vision, comprising the following modules: a printed matter sample selection module, used to select samples from batches of printed matters produced in the same batch using the same raw materials.

[0007] The imaging test module is used to perform imaging tests on selected printed product samples, thereby determining appropriate imaging parameters for batches of printed products.

[0008] The defect background feature extraction module is used to collect the apparent image of the printed sample after adjusting the imaging device based on appropriate imaging parameters, and extract the defect background features from the apparent image.

[0009] The image preprocessing test module is used to perform image preprocessing tests based on defect background features in the apparent image of printed samples, thereby determining the appropriate image preprocessing sequence after imaging of batch printed products.

[0010] The multi-source image acquisition module is used to acquire the apparent image of the batch of printed products using the appropriate imaging parameters of the batch of printed products, and simultaneously acquire the infrared image to obtain the apparent image and infrared image of each printed product.

[0011] The image preprocessing module is used to perform image preprocessing on the apparent images of batch printed products according to an appropriate image preprocessing sequence.

[0012] The printing quality inspection module is used to perform surface defect inspection on the surface image of each printed product in a batch of printed products after image preprocessing, and to perform internal structural defect inspection on the infrared image of each printed product.

[0013] The printing quality assessment module is used to assess the printing quality based on the detection results of the surface defects and internal structural defects of each printed product in a batch of printed products.

[0014] Combining all the above technical solutions, the present invention achieves the following positive effects: 1. By selecting samples from batches of printed cigarette packs, conducting imaging tests based on the samples, and determining appropriate imaging parameters based on the test results, the present invention then captures surface images of the batches of printed cigarette packs according to these parameters. This process maximizes the quality of surface image acquisition, providing reliable data support for surface defect identification, and significantly improving the reliability and accuracy of the inspection system.

[0015] 2. After conducting an imaging test on the selected cigarette box printed product samples, the present invention adds defect background feature extraction of the sample imaging results, and conducts an image preprocessing test to determine the optimal image preprocessing sequence. This process can provide an efficient processing solution for the apparent image preprocessing of batch cigarette box printed products, ensuring that the defect features in the image are more prominent, thereby enhancing the efficiency and accuracy of apparent defect recognition of batch cigarette box printed products.

[0016] 3. This invention uses appropriate imaging parameters to simultaneously capture surface images of batches of printed cigarette packs while also simultaneously capturing infrared images. By combining surface defects identified through surface images with internal structural defects identified through infrared images, the system enables comprehensive print quality assessment and complete defect identification. This approach not only improves the accuracy and completeness of defect detection, significantly reduces the risk of missed and false detections, but also significantly enhances the efficiency and reliability of print quality control. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The present invention is further described with reference to the accompanying drawings. However, the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of system module connections of the present invention.

[0019] Figure 2 Schematic diagram of the operation of the imaging test in the present invention.

[0020] Figure 3 This is an operation diagram for batch printing quality inspection and evaluation in the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0022] See also Figure 1 As shown, the present invention provides a printed product printing quality detection and analysis system based on machine vision, including a printed product sample selection module, an imaging test module, a defect background feature extraction module, an image preprocessing test module, a multi-source image acquisition module, an image preprocessing module, a printing quality detection module and a printing quality evaluation module, wherein the printed product sample selection module is connected to the imaging test module, the imaging test module is connected to the defect background feature extraction module, the defect background feature extraction module is connected to the image preprocessing test module, the imaging test module is connected to the multi-source image acquisition module, the multi-source image acquisition module and the image preprocessing test module are both connected to the image preprocessing module, the image preprocessing module is connected to the printing quality detection module, and the printing quality detection module is connected to the printing quality evaluation module.

[0023] The printed product sample selection module is used to select samples from batches of printed products produced in the same batch using the same raw materials.

[0024] In the implementation of the above scheme, the specific operation of sample selection is as follows: the total number of batch prints is counted, and each print is numbered, wherein the numbering can be performed according to the production order of the prints.

[0025] The number of samples is determined from the total number according to the predetermined sample selection ratio.

[0026] For example, the sample selection ratio may be 10%, and when the number of printed products in a batch is 500, the number of samples determined is 50.

[0027] Based on the determined number of samples, the number intervals between the batches of printed products are calculated, and the samples are selected according to the intervals.

[0028] In a further example, when the number of samples is 50, the numbering interval is 500 divided by 50, which equals 10. That is, starting from the print numbered 1, one print is selected as a sample every 10 numbers, and the sample numbers finally selected are 1, 11, 21, 31...491, for a total of 50 prints.

[0029] It is important to understand that by selecting samples from batches of printed materials at numbered intervals, the representativeness of the samples can be ensured, bias can be avoided, and the accuracy and reliability of the test results can be improved. In addition, this method can ensure that the samples come from different locations throughout the entire production batch, avoiding local bias caused by selecting samples from only a certain part. This helps to ensure that the samples are evenly distributed and cover the entire production batch. It can more comprehensively reflect the overall quality status of the batch of printed materials, provide representative imaging samples for subsequent imaging tests, and avoid the problems of repeated tests or oversampling. This can ensure the authenticity of the imaging test results. It can also reduce test time and resource consumption while ensuring the accuracy of the test results, thereby improving test efficiency.

[0030] See also Figure 2 As shown, the imaging test module is used to perform an imaging test on the selected printed product samples, thereby determining suitable imaging parameters for batch printed products.

[0031] It is important to know that when conducting quality inspections on printed products, it is necessary to use imaging equipment to capture images of the printed product surface, where the imaging equipment is a visible light imaging device, specifically a high-definition camera, and identify defects (such as scratches, stains, etc.) from the captured apparent image. However, since the raw materials used to produce printed products have different properties such as smoothness and color, these properties will have a significant impact on the image acquisition effect. For example, a rougher paper surface may produce more scattered light, resulting in shadows, reflections, or blurring during image acquisition; different colors of paper have different absorption and reflection characteristics for light. Light-colored paper reflects more light, which can easily lead to overexposure; while dark-colored paper absorbs more light, which can easily lead to underexposure. Therefore, when capturing the apparent image of a printed product, it is necessary to determine the appropriate imaging parameters for the printed product in the corresponding production batch through imaging tests to ensure the best imaging effect.

[0032] Applied to the above operation, the imaging test is specifically implemented as follows: determining the imaging parameter variables to be tested, thereby forming an imaging test matrix with imaging parameter combinations, and counting the number of imaging test groups.

[0033] In the above example, the imaging parameter variables include exposure time, polarization angle, and gain. Based on the specifications of the imaging device, you can obtain the control range of each imaging parameter variable from the user manual. These control ranges can then be divided into multiple test values ​​at equal intervals. For example, if the exposure time control range is [X1, X2], this can be divided into multiple test values ​​at equal intervals; if the polarization angle control range is [Y1, Y2], this can be divided into multiple test values; and if the gain control range is [Z1, Z2], this can be divided into multiple test values.

[0034] It is important to understand that exposure time, polarization angle, and gain were chosen as imaging parameter variables because they significantly impact image quality during acquisition. Each parameter directly affects image brightness, contrast, detail clarity, and noise levels, which in turn impacts the accuracy and reliability of defect detection.

[0035] Specifically, exposure time refers to the length of time the sensor is exposed to light when capturing an image. Longer exposure times allow the sensor to receive more light, resulting in a brighter image. Conversely, shorter exposure times result in darker images. Proper exposure time ensures moderate image brightness, avoiding overexposure or underexposure.

[0036] Polarization refers to the process of adjusting the vibration direction of light through a polarizing filter or light source, so that it propagates in a specific direction. Printed surfaces, especially smooth or glossy materials, can produce reflections and glare that interfere with image quality. Polarized light can effectively reduce these reflections, enhancing image detail.

[0037] Gain refers to the sensor's signal amplification factor, which determines the image brightness and noise level. Higher gain results in brighter images, while lower gain results in darker images. Proper gain settings ensure moderate image brightness, avoiding overly bright or dark images.

[0038] From the experimental values ​​of each imaging parameter variable, one experimental value is selected in turn to form an imaging test matrix. For example, assuming that there are 3 experimental values ​​for exposure time, polarization light, and gain, the imaging test matrix is ​​3×3×3, that is, 27 imaging test groups.

[0039] Divide the selected printed material samples evenly into corresponding groups based on the number of imaging test groups. This ensures that after each imaging test group is tested, there are enough printed material samples to present the imaging test results, avoiding the lack of representativeness and biased results caused by a single printed material sample.

[0040] When performing the imaging test of each imaging test group, the imaging environment state is controlled to be consistent, and after the test, the apparent image of the corresponding printed product sample of each imaging test group is obtained.

[0041] It is important to note that when conducting imaging tests, the imaging environment must be kept consistent, specifically by ensuring stable control of conditions such as light intensity and ambient temperature. This can avoid interference from external factors and prevent fluctuations in test results and data deviations caused by environmental changes.

[0042] Further applied to the above operation, the appropriate imaging parameters for batch prints are determined as follows: visual effect indicators are extracted from the apparent images of the corresponding print samples of each imaging test group. For example, the visual effect indicators may include edge intensity, contrast and color depth, thereby performing image visual effect scoring for each print sample.

[0043] It's important to note that edge strength indicates the clarity of image boundaries and details. High edge strength indicates distinct outlines and sharp details, improving image readability and visual clarity. Contrast refers to the difference between the brightest and darkest parts of an image; a high-contrast image shows a clear distinction between bright and dark areas. Color depth refers to the number of colors that can be displayed in an image, typically expressed in bits (e.g., 24-bit color). The higher the color depth, the richer the image's color gradation. Using edge strength, contrast, and color depth as visual quality indicators allows for a comprehensive assessment of an image's visual quality.

[0044] The above image visual effect scoring formula is Where p represents the image visual effect rating value, ES, CR, and CD represent edge strength, contrast, and color depth, respectively; ES0, CR0, and CD0 represent the ideal edge strength, ideal contrast, and ideal color depth of the imaging device, respectively; and R represents the rating system, which can be one point, ten points, or one hundred points. When R is one point, R = 1; when R is ten points, R = 10.

[0045] It should be added that the ideal edge intensity, ideal contrast, and ideal color depth of the imaging device mentioned above can be extracted from the operating instructions of the imaging device, and the maximum values ​​that the imaging device can achieve can be used as ideal values. From the above image visual effect scoring formula, it can be seen that the closer the visual effect indicators of the apparent image of the printed sample are to the ideal values, the greater the image visual effect scoring value.

[0046] The image visual effect score of each printed sample in each imaging test group was compared with the full score to calculate the imaging effect coefficient corresponding to each imaging test group. The specific calculation formula is: Where Q represents the imaging effect coefficient, p i represents the image visual effect score of the i-th printed sample in the imaging test group, i represents the number of the printed sample in the imaging test group, i = 1, 2, ..., n, n represents the number of printed samples in the imaging test group, p0 represents the full score, e represents the natural constant, p i max, p i min represents the maximum and minimum values ​​of the visual effect scores of the printed samples in the imaging test group, respectively, and η allow represents the preset allowed fluctuation of the image visual effect score. For example, η allow=0.3, where p′ represents the visual effect score of the set standard image.

[0047] It should be pointed out that when the scoring system is one point, the full score is 1 and the standard score is 0.7; when the scoring system is ten points, the full score is 10 and the standard score is 7; when the scoring system is one hundred points, the full score is 100 and the standard score is 70.

[0048] When evaluating the imaging effect coefficient, it is taken into account that the image visual effect scores of the printed samples in each imaging test group may fluctuate. When the fluctuation is small, it means that the image visual effects of each printed sample are relatively consistent, and the imaging system performs stably on different samples. At this time, comparing the image visual effect score of each printed sample with the full score can effectively evaluate the overall performance of the imaging system. If the scores of all samples are close to the full score, the imaging effect is better. When the fluctuation is large, it means that there are significant differences in the image visual effects of each printed sample, which may be caused by experimental errors. At this time, comparing the image visual effect score of each printed sample with the set standard score can more reasonably evaluate the minimum qualification standard of the imaging system. If the scores of most printed samples can reach the standard score, it means that the imaging system can meet the basic quality requirements in most cases.

[0049] The imaging effect coefficient corresponding to each imaging test group is compared with the set standard threshold. For example, the standard threshold is 0.8. If the imaging effect coefficient corresponding to a certain imaging test group reaches the standard threshold, the imaging test group is regarded as a valid imaging test group.

[0050] The number of valid imaging test groups is counted. If there is only one valid imaging test group, the imaging parameter combination corresponding to the imaging test group is used as the appropriate imaging parameter for batch printing. If there is more than one valid imaging test group, the imaging parameter combination corresponding to the valid imaging test group with the maximum imaging effect coefficient is used as the appropriate imaging parameter for batch printing.

[0051] If no imaging test group that meets the threshold is screened out, the imaging test groups corresponding to the maximum imaging effect coefficient and the second largest imaging effect coefficient are extracted from the imaging effect coefficients corresponding to each imaging test group and recorded as the first imaging test group and the second imaging test group respectively.

[0052] Imaging parameter refinement is performed between the imaging parameter combinations corresponding to the first imaging test group and the second imaging test group to form a new imaging test group.

[0053] It's important to note that parameter refinement involves fine-tuning the parameters between two imaging test groups with the best performance to find the optimal imaging parameter combination. This approach, based on the principle of gradient optimization, gradually approaches the optimal parameters and has a scientific basis. By refining the parameters between the maximum and second-largest imaging effect coefficients, a blind search from scratch can be avoided, improving optimization efficiency and accuracy.

[0054] Imaging tests are continued under the new imaging test groups to obtain the apparent images of the printed samples corresponding to each imaging test group. Similarly, imaging effect coefficient analysis is performed and compared with the set standard threshold, and then the imaging test group that meets the standard threshold is screened out, so that the imaging parameter combination corresponding to the imaging test group is used as the appropriate imaging parameter for batch prints.

[0055] The present invention determines appropriate imaging parameters based on imaging test results through an initial screening process followed by parameter refinement. The initial screening phase involves designing an imaging test matrix and conducting comprehensive parameter testing to ensure coverage of all possible imaging parameter combinations. This provides a solid foundation for subsequent optimization. Parameter refinement in the subsequent optimization phase allows focus on high-performing parameter combinations, avoiding overexploration of ineffective parameters and saving time and resources.

[0056] The defect background feature extraction module is used to collect the apparent image of the printed product sample after adjusting the imaging device based on appropriate imaging parameters, and extract the defect background feature from the apparent image.

[0057] It should be added that after the apparent image of the printed sample is acquired, the image needs to be preprocessed before defect identification. This is because in order to achieve accurate defect segmentation during the defect identification process, the defect area must be effectively separated from the background. To this end, it is necessary to extract the defect background features in the apparent image. These features include noise features, color features, geometric shape features, etc., which significantly affect the extraction and segmentation of defects. Since the transformations and adjustments introduced into the image by each preprocessing step are interdependent, these steps may enhance or weaken certain features and affect the effects of subsequent steps. This makes different image preprocessing step sequences have different effects on defect extraction and segmentation results. Therefore, it is necessary to determine the optimal preprocessing sequence through image preprocessing experiments to ensure the accuracy and reliability of defect identification.

[0058] The image preprocessing test module is used to perform an image preprocessing test based on defect background features in the apparent image of a printed product sample, thereby determining an appropriate image preprocessing sequence after imaging of a batch of printed products.

[0059] Preferably, the image preprocessing test refers to the following process: the required image preprocessing method is determined based on the defect background features extracted from the apparent image of the printed sample. Specifically, when the noise features extracted from the defect background features are noisy, the required image preprocessing method is denoising. When the color features extracted from the defect background features have color differences, the required image preprocessing method is color segmentation, which can help separate the defect area. When the geometric shape features extracted from the defect background features are relatively complex, the required image preprocessing method is morphological operation, which can help simplify the image structure and highlight the defect features. Common morphological operations include: dilation and erosion, opening and closing operations, skeleton extraction, etc.

[0060] The required image preprocessing methods are sequentially combined to form several image preprocessing sequence groups.

[0061] In the example of the above operation, when the required image preprocessing method is denoising, color segmentation and morphological operation, the image preprocessing order groups formed at this time are the first group: first perform denoising, then color segmentation and finally perform morphological operation; the second group: first perform color segmentation, then perform denoising and finally perform morphological operation; the third group: first perform color segmentation, then perform morphological operation and finally perform denoising; the fourth group: first perform color segmentation, then perform morphological operation and finally perform denoising; the fifth group: first perform color segmentation, then perform denoising and finally perform morphological operation.

[0062] The apparent image of the printed sample was preprocessed under each image preprocessing sequence group, and the preprocessing time and defect detection related parameters under each sequence group were recorded.

[0063] The preprocessing time mentioned above refers to the time required to complete a series of preprocessing steps, which directly affects the overall defect detection efficiency. Shorter preprocessing time means that image processing can be completed in a shorter time, improving the efficiency of image quality analysis. This is especially important when quality testing batches of printed products, as time cost is a key consideration.

[0064] The above-mentioned defect detection related parameters can be defect detection rate and false detection rate. The defect detection rate refers to the proportion of successfully detected defects among actual defects, and the false detection rate refers to the proportion of all non-defective areas that are mistakenly identified as defective areas.

[0065] Further preferably, the appropriate image preprocessing sequence after batch printing is imaged is determined as follows: defect detection effect scoring is performed based on the defect detection associated parameters of the printed sample appearance image under each image preprocessing sequence group. Specifically, the defect detection effect scoring formula is: Where λ and δ represent the defect detection rate and false detection rate, respectively.

[0066] The preprocessing time and defect detection effect score of the printed sample appearance image in each image preprocessing sequence group are statistically analyzed to determine the presentation value of the printed sample appearance image in each image preprocessing sequence group. The specific statistical formula is: Where PV represents the presentation value, t represents the preprocessing time, s represents the defect detection effect score, s0 represents the full score of the defect detection effect, t 总 It represents the cumulative preprocessing time of the apparent image of the printed sample under each image preprocessing sequence group.

[0067] The presentation values ​​of the printed matter sample appearance image in each image preprocessing sequence group are compared, and the sequence group with the largest presentation value is selected as the preferred image preprocessing sequence group for the printed matter sample appearance image.

[0068] This paper conducts image preprocessing experiments, recording the preprocessing time, defect detection rate, and false positive rate for each image preprocessing sequence. This allows us to find the optimal balance between accuracy and efficiency. Certain preprocessing sequences may improve the defect detection rate, but also increase the false positive rate or preprocessing time; while other sequences may shorten the preprocessing time but reduce the defect detection rate. Therefore, a comprehensive evaluation of these parameters can help determine the optimal preprocessing sequence, ensuring maximum efficiency while meeting accuracy requirements.

[0069] The preferred image preprocessing sequence groups of the apparent image of each printed sample are compared, and the preferred image preprocessing sequence group with the highest occurrence frequency is selected, and the image preprocessing sequence in this group is used as the appropriate image preprocessing sequence after imaging of batch printed products.

[0070] It should be noted that because batches of printed products are produced using the same raw materials, the defect background features in the apparent images obtained using appropriate imaging parameters are largely similar. Therefore, the preferred image preprocessing sequence group for the apparent image of each printed product sample should have a certain degree of universality. By comparing the preferred image preprocessing sequence groups for each printed product sample, the sequence group with the highest frequency of occurrence is selected and used as the standard image preprocessing sequence for batch print imaging. This ensures consistency and stability in the imaging and defect detection processes for batch prints.

[0071] The multi-source image acquisition module is used to acquire the apparent image of the batch of printed products using appropriate imaging parameters of the batch of printed products, and simultaneously acquire infrared images to obtain the apparent image and infrared image of each printed product.

[0072] The image preprocessing module is used to perform image preprocessing on the apparent images of batch printed products according to an appropriate image preprocessing sequence.

[0073] It should be noted that since the surface image is collected using visible light imaging equipment, the raw material of the printed matter will significantly affect the visual effect of the image. Therefore, in order to ensure the efficiency and high quality of the surface image, imaging tests and image preprocessing tests must be conducted to optimize imaging parameters and preprocessing steps to improve the accuracy and reliability of defect detection. In contrast, infrared images are collected using infrared imaging equipment, which uses infrared radiation to penetrate the surface of the printed matter to form an image. Since infrared imaging mainly relies on the thermal radiation properties of the object rather than the reflective properties of visible light, the raw material of the printed matter has little effect on the visual effect of the infrared image. Therefore, infrared images generally do not require complex imaging tests, but imaging tests are also an option.

[0074] See also Figure 3 As shown, the printing quality detection module is used to perform surface defect detection on the surface image of each printed product in a batch of printed products after image preprocessing, and to perform internal structural defect detection on the infrared image of each printed product.

[0075] It should be noted that before using surface images and infrared images for defect detection, they must be registered. This is because surface images and infrared images typically come from different imaging modalities and have different physical properties and imaging mechanisms. These differences can lead to spatial inconsistencies between the two images, thus affecting subsequent fusion, comparison, and analysis.

[0076] The above-mentioned surface inspection is as follows: apparent defects are captured from the apparent image of the printed product, and the location and characteristic information of each apparent defect are marked. The characteristic information includes but is not limited to defect shape, defect color, defect texture, defect size, etc., among which defect shape, defect color, and defect texture can be used to distinguish defect types, such as scratches, stains, and ink diffusion.

[0077] Internal structural defect detection involves capturing structural defects from infrared images of printed materials and annotating the location and characteristic information of each structural defect. This characteristic information includes defect thickness and depth. Defect thickness refers to the maximum or average thickness of the defect in a particular direction. A greater thickness generally indicates a more severe defect. Defect depth refers to the depth of the defect within the material.

[0078] The printing quality assessment module is used to perform printing quality assessment based on the detection results of surface defects and internal structural defects of each printed product in a batch of printed products. Specifically, the printing quality assessment refers to the following process: the locations of each surface defect and each structural defect of the printed product are compared to determine whether they are the same defect area, thereby merging the defect areas in the printed product.

[0079] In the manner in which the above operation can be implemented, the following process is used to determine whether it is the same defect area: the boundary contours of the apparent defects and structural defects at each location of the printed product are extracted separately to calculate their overlap. Specifically, the overlap can be obtained by dividing the area of ​​the overlapping area by the area of ​​the apparent defect boundary contour, and then comparing it with a preset overlap threshold. For example, the overlap threshold is 90%. If the overlap between the boundary contour of a certain apparent defect and the boundary contour of a certain structural defect reaches or exceeds the overlap threshold, it is determined that the two defects belong to the same defect area.

[0080] The apparent defect characteristics and structural defect characteristics of each defect area after merging are integrated to evaluate the quality coefficient of each printed product.

[0081] Specifically, when only superficial defects exist in a defective area, the defect type (such as scratches, stains, etc.) and defect size (such as length, area, etc.) in the superficial defect information are used to quantify the defect degree to obtain the defect degree value of the defective area.

[0082] For example, defect severity can be quantified by introducing impact factors for different defect types. The specific steps are as follows: First, an impact factor is set for each defect type, which reflects the degree of impact of that defect on product quality and performance. Then, the size of each defect is measured, its area is calculated, and this is divided by the total surface area of ​​the printed product to obtain the defect percentage. Finally, the impact factor corresponding to the defect type is multiplied by the defect percentage to obtain the final defect severity value.

[0083] When only internal structural defects exist in a defective area, the defect degree is quantified using the internal structural defect information (such as defect thickness and defect depth) to obtain the defect degree value of the defective area.

[0084] When a defective area has both surface defects and internal structural defects, the two defects are quantified separately, and the defect degree values ​​of the two are added together to obtain the total defect degree value of the defective area.

[0085] Finally, the quality coefficient of each printed product is calculated by accumulating the defect degree values ​​of all defective areas in each printed product.

[0086] The quality coefficient of each printed product is compared with the qualified quality coefficient of the corresponding batch of printed products, where the qualified quality coefficient can be extracted from the production documents of the batch of printed products, thereby calculating the quality compliance rate of the batch of printed products. Specifically, the number of printed products that reach the qualified quality coefficient can be calculated divided by the total number of batch printed products.

[0087] The above content is merely an example and explanation of the structure of the present invention. Those skilled in the art may make various modifications or additions to the described specific embodiments or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the present invention, they should all fall within the scope of protection of the present invention.

Claims

1. A printed matter printing quality detection and analysis system based on machine vision, characterized in that: Includes the following modules: The printed product sample selection module is used to select samples from batches of printed products produced using the same raw materials and the same batch; An imaging test module is used to perform imaging tests on selected printed product samples to determine suitable imaging parameters for batches of printed products; The imaging test on the selected printed product sample is carried out as follows: Determine the imaging parameter variables that need to be tested, thereby forming an imaging test matrix with imaging parameter combinations, and count the number of imaging test groups; According to the number of imaging test groups, the selected printed samples were evenly divided into corresponding groups; When performing imaging tests of each imaging test group, the imaging environment state is controlled to be consistent, and after the test, the apparent image of the corresponding printed product sample of each imaging test group is obtained; A defect background feature extraction module is used to collect an apparent image of a printed product sample after adjusting an imaging device based on appropriate imaging parameters, and to extract defect background features from the apparent image; An image preprocessing test module is used to perform image preprocessing tests based on defect background features in the apparent image of printed product samples, thereby determining the appropriate image preprocessing sequence after imaging of batch printed products; A multi-source image acquisition module is used to acquire the apparent image of the batch of printed products using appropriate imaging parameters of the batch of printed products, and simultaneously acquire infrared images to obtain the apparent image and infrared image of each printed product; An image preprocessing module, for performing image preprocessing on the apparent images of batch printed products according to an appropriate image preprocessing sequence; The printing quality inspection module is used to detect surface defects on the surface images of each printed product in a batch of printed products after image preprocessing, and to detect internal structural defects on the infrared images of each printed product; The printing quality assessment module is used to assess the printing quality based on the detection results of the surface defects and internal structural defects of each printed product in a batch of printed products.

2. The machine vision-based printed matter printing quality detection and analysis system according to claim 1, characterized in that: The sample selection process is as follows: Count the total number of printed materials in a batch and number each printed material; Determine the number of samples from the total number according to the predetermined sample selection ratio; Based on the determined number of samples, the number intervals between the batches of printed products are calculated, and the samples are selected according to the intervals.

3. The machine vision-based printed matter printing quality detection and analysis system according to claim 1, characterized in that: The determination of suitable imaging parameters for batch prints is carried out as follows: Extract visual effect indicators from the apparent images of the corresponding printed product samples of each imaging test group, and thereby perform image visual effect scoring of each printed product sample; The image visual effect score of each printed sample in each imaging test group was compared with the full score to calculate the imaging effect coefficient corresponding to each imaging test group. The specific calculation formula is: , where represents the imaging effect coefficient, Indicates the first Image visual effect rating of printed samples, Indicates the print sample number in the imaging test group, , Indicates the number of printed samples in the imaging test group, Indicates a full score. represents a natural constant, 、 Respectively represent the maximum and minimum values ​​of the visual effect scores of the corresponding images of each printed sample in the imaging test group, Indicates the allowed fluctuation of the preset image visual effect score, where , Indicates the visual effect score of the set target image; Compare the imaging effect coefficient corresponding to each imaging test group with the set standard threshold. If the imaging effect coefficient corresponding to a certain imaging test group reaches the standard threshold, then the imaging test group is regarded as a valid imaging test group. The number of valid imaging test groups is counted. If there is only one valid imaging test group, the imaging parameter combination corresponding to the imaging test group is used as the appropriate imaging parameter for batch printing. If there is more than one valid imaging test group, the imaging parameter combination corresponding to the valid imaging test group with the maximum imaging effect coefficient is used as the appropriate imaging parameter for batch printing.

4. The machine vision-based printed matter printing quality detection and analysis system according to claim 3, characterized in that: The method of determining suitable imaging parameters for batch printed products further comprises the following steps: If no imaging test group that meets the threshold is screened out, the imaging test groups corresponding to the maximum imaging effect coefficient and the second largest imaging effect coefficient are extracted from the imaging effect coefficients corresponding to each imaging test group and recorded as the first imaging test group and the second imaging test group respectively; Refining imaging parameters between imaging parameter combinations corresponding to the first imaging test group and the second imaging test group to form a new imaging test group; Imaging tests are continued under the new imaging test groups to obtain the apparent images of the printed samples corresponding to each imaging test group. Similarly, imaging effect coefficient analysis is performed and compared with the set standard threshold, and then the imaging test group that meets the standard threshold is screened out, so that the imaging parameter combination corresponding to the imaging test group is used as the appropriate imaging parameter for batch prints.

5. The machine vision-based printed matter printing quality detection and analysis system according to claim 1, characterized in that: The image preprocessing test is as follows: Determine the required image preprocessing method based on the defect background features extracted from the surface image of the printed sample; Sequentially combining the required image preprocessing methods to form a number of image preprocessing sequence groups; The apparent image of the printed sample was preprocessed under each image preprocessing sequence group, and the preprocessing time and defect detection related parameters under each sequence group were recorded.

6. The machine vision-based printed matter printing quality detection and analysis system according to claim 5, characterized in that: The process of determining the appropriate image preprocessing sequence after batch printing is performed is as follows: The defect detection effect is scored based on the defect detection correlation parameters of the printed sample surface image under each image preprocessing sequence group; The preprocessing time and defect detection effect score of the printed sample appearance image in each image preprocessing sequence group are statistically analyzed to determine the presentation value of the printed sample appearance image in each image preprocessing sequence group. The specific statistical formula is: , where Indicates the value of presentation. Indicates the preprocessing time. Indicates the defect detection effect score, Indicates the full score of defect detection effect, It represents the cumulative preprocessing time of the apparent image of the printed sample under each image preprocessing sequence group; Comparing the presentation values ​​of the printed matter sample's apparent image in each image preprocessing sequence group, and selecting the sequence group with the greatest presentation value as the preferred image preprocessing sequence group for the printed matter sample's apparent image; The preferred image preprocessing sequence groups of the apparent image of each printed sample are compared, and the preferred image preprocessing sequence group with the highest occurrence frequency is selected, and the image preprocessing sequence in this group is used as the appropriate image preprocessing sequence after imaging of batch printed products.

7. The machine vision-based printed matter printing quality detection and analysis system according to claim 1, characterized in that: The surface defect detection refers to the following process: Capture surface defects from the surface image of the printed product and mark the location and characteristic information of each surface defect; The internal structural defect detection is as follows: Structural defects are captured from infrared images of printed materials, and the location and characteristic information of each structural defect are marked.

8. The machine vision-based printed matter printing quality detection and analysis system according to claim 7, characterized in that: The printing quality evaluation is as follows: Comparing the locations of the apparent defects and the structural defects of the printed product to determine whether they are the same defect area, thereby merging the defect areas in the printed product; The apparent defect characteristics and structural defect characteristics of each defect area after merging are integrated to evaluate the quality coefficient of each printed product; The quality coefficient of each printed product is compared with the qualified quality coefficient of the corresponding batch of printed products, and the quality compliance rate of the batch of printed products is calculated.

9. The machine vision-based printed matter printing quality detection and analysis system according to claim 8, characterized in that: The process of determining whether the defect area is the same is as follows: The boundary contours of the apparent defects and structural defects at each location on the printed product are extracted separately to calculate their overlap, and then compared with the preset overlap threshold. If the overlap between the boundary contour of a certain apparent defect and the boundary contour of a certain structural defect reaches or exceeds the overlap threshold, it is determined that the two defects belong to the same defect area.

Images