A printing anomaly detection and early warning system
By analyzing the grayscale and gradient information of pixels in printed images, calculating the degree of printing boundary and adaptively adjusting the size of superpixel blocks, the problem of difficult detection of ink spot defect areas is solved, the detection accuracy is improved and the loss of normal pixels is reduced.
Patent Information
- Application Number
- CN202410574939.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-10
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-05-10
AI Technical Summary
In existing technologies, ink spot defect areas are similar to printed text or patterns, causing the CA saliency detection algorithm to fail to effectively detect ink spot defect areas, reducing detection accuracy and losing normal pixels.
By analyzing the correlation and dense distribution of grayscale and gradient information among different pixels, the printing boundary degree of each pixel is calculated, the superpixel block size is adaptively adjusted, and anomaly detection is performed using the CA saliency detection algorithm.
It improves the detection accuracy of ink spot defect areas, reduces the interference of normal areas on pixel identification, and reduces the loss of normal pixels.
Smart Images

Figure CN118396967B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image processing, and in particular to a printing anomaly detection and early warning system. Background Art
[0002] During the printing process, the ink tank inside the printing device will be affected by the temperature, humidity and ink concentration of the surrounding environment when the ink is discharged, which may cause ink spots on the printed products and quality problems of the printed products. Therefore, it is necessary to perform quality inspection on the printed images.
[0003] The existing technology usually uses a CA (Context-Aware) saliency detection algorithm to detect ink spot defect areas in printed images. However, since some ink spots are similar to some printed text and patterns, some pixels belonging to the ink spot defect area will have saliency values similar to some normal pixels. Therefore, the ink spot defect area cannot be effectively detected, which reduces the accuracy of detecting the ink spot defect area and causes some normal pixels to be lost. Summary of the invention
[0004] The present invention provides a printing anomaly detection and early warning system to solve the existing problem: since some ink spots are similar to some printed text and patterns, some pixel points belonging to the ink spot defect area will have significant values similar to some normal pixel points, and the ink spot defect area cannot be effectively detected.
[0005] A printing anomaly detection and early warning system of the present invention adopts the following technical solution:
[0006] Includes the following modules:
[0007] A printed matter grayscale image acquisition module, used for acquiring printed matter grayscale images of a plurality of printed matters;
[0008] The ink spot defect pixel recognition module is used to obtain the printing boundary degree of each pixel point according to the correlation between the grayscale and gradient information of different pixels and the dense distribution of the corresponding positions; all pixels are divided according to the printing boundary degree of each pixel point to obtain a number of ink spot defect pixels;
[0009] The anomaly detection module is used to adaptively adjust the size of the super pixel block according to the printing boundary degree of each ink spot defect pixel point and perform printing anomaly detection.
[0010] Preferably, the printing boundary degree of each pixel is obtained according to the correlation between the grayscale and gradient information of different pixels and the dense distribution of the corresponding positions, and the specific method includes:
[0011] For any printed grayscale image, the Sobel operator is used to obtain the gradient amplitude and gradient direction degree of all pixels in the printed grayscale image; a window size TL1 is preset, and any pixel is recorded as the target pixel. The window area with a window size of TL1×TL1 centered on the target pixel is used as the reference neighborhood area of the target pixel; each pixel in the reference neighborhood area of the target pixel, except the target pixel, is recorded as the reference pixel of the target pixel;
[0012] For the lth reference pixel, according to the difference in gray value, gradient amplitude and gradient direction degree between the target pixel and the lth reference pixel, the printing feature same-direction correlation coefficient between the target pixel and the lth reference pixel is obtained;
[0013] According to the printing feature unidirectional correlation coefficient and the corresponding distance between the target pixel and the lth reference pixel, the printing feature correlation factor between the target pixel and the lth reference pixel is obtained; the printing feature correlation factors between the target pixel and all the reference pixels are linearly normalized, and each normalized printing correlation factor is recorded as the printing correlation degree;
[0014] According to the distribution law of the printing feature correlation degree between the target pixel and all the reference pixels, the regional feature correlation degree of the target pixel is obtained; the regional feature correlation degree of each pixel is obtained; and the sequence of all the pixels arranged in descending order according to the regional feature correlation degree is recorded as the feature correlation pixel sequence;
[0015] According to the degree of feature association and distribution aggregation between adjacent pixels in the feature-associated pixel sequence, a printing boundary weight coefficient of each pixel in the feature-associated pixel sequence is obtained;
[0016] According to the difference comparison between the regional feature correlation degree between the i1th pixel point on the feature correlation pixel point sequence and the pixels at both ends of the feature correlation pixel point sequence, the reasonableness of the selection of the regional feature correlation degree of the i1th pixel point is obtained;
[0017] According to the difference distribution of the regional feature association degree between adjacent pixels in the feature association pixel sequence and the printing boundary weight coefficient of the i1th pixel in the feature association pixel sequence, the reasonable degree of printing feature change of the i1th pixel is obtained;
[0018] The product of the reasonable degree of selection of the regional feature association degree of the i1-th pixel point and the reasonable degree of change of the printing feature of the i1-th pixel point is taken as the printing boundary degree of the i1-th pixel point.
[0019] Preferably, the method of obtaining the printing feature same-direction correlation coefficient between the target pixel and the lth reference pixel according to the difference in grayscale value, gradient amplitude and gradient direction degree between the target pixel and the lth reference pixel includes the following specific methods:
[0020] ε1=|HH l |×|A1-A1 l |×|θ-θ l |
[0021] In the formula, ε l represents the printing feature unidirectional correlation coefficient between the target pixel and the lth reference pixel; H represents the gray value of the target pixel; H l A1 represents the gray value of the lth reference pixel; A2 represents the gradient amplitude of the target pixel; A3 represents the gray value of the lth reference pixel; A4 represents the gray value of the lth reference pixel; A5 represents the gray value of the lth reference pixel; A6 represents the gray value of the lth reference pixel; A7 represents the gray value of the lth reference pixel; A8 represents the gray value of the lth reference pixel; A9 represents the gray value of the lth reference pixel; A1 l represents the gradient magnitude of the lth reference pixel; θ represents the gradient direction degree of the target pixel; θ l Represents the gradient direction degree of the lth reference pixel; || represents the absolute value.
[0022] Preferably, the printing feature correlation factor between the target pixel and the lth reference pixel is obtained according to the printing feature unidirectional correlation coefficient between the target pixel and the lth reference pixel and the corresponding distance, and the specific method includes:
[0023] The ratio of the printing feature unidirectional correlation coefficient between the target pixel and the lth reference pixel to the Euclidean distance between the target pixel and the lth reference pixel is used as the printing feature correlation factor between the target pixel and the lth reference pixel.
[0024] Preferably, the method of obtaining the regional feature correlation degree of the target pixel point according to the distribution law of the printing feature correlation degree between the target pixel point and all reference pixels includes the following specific methods:
[0025] The standard deviation of the printing feature association degree between the target pixel and all reference pixels is used as the regional feature association degree of the target pixel.
[0026] Preferably, the method of obtaining the printing boundary weight coefficient of each pixel in the feature-associated pixel sequence according to the feature association degree and distribution aggregation between adjacent pixels in the feature-associated pixel sequence includes the following specific methods:
[0027] For any two adjacent pixels in the feature-associated pixel sequence, the difference in the degree of feature association between the first pixel and the second pixel is recorded as the first difference, the Euclidean distance between the first pixel and the second pixel is recorded as the first distance, and the ratio of the first difference to the first distance is recorded as the printing boundary weight coefficient of the first pixel.
[0028] Preferably, the method of obtaining the reasonableness of selecting the regional feature association degree of the i1th pixel point according to the comparison of the difference between the regional feature association degree of the i1th pixel point on the feature-associated pixel point sequence and the pixels at both ends of the feature-associated pixel point sequence includes the following specific methods:
[0029]
[0030] In the formula, ε1 i1 represents the reasonableness of the selection of the regional feature association degree of the i1th pixel point; Q1 represents the regional feature association degree of the first pixel point in the feature association pixel point sequence; Q i1 It represents the regional feature correlation degree of the i1th pixel; Q2 represents the regional feature correlation degree of the last pixel in the feature-correlated pixel sequence; exp() represents an exponential function with a natural constant as the base; || represents taking the absolute value.
[0031] Preferably, the specific method of obtaining the reasonable degree of printing feature change of the i1th pixel point according to the difference distribution of regional feature association degrees between adjacent pixels in the feature-associated pixel point sequence and the printing boundary weight coefficient of the i1th pixel point in the feature-associated pixel point sequence is as follows:
[0032]
[0033] In the formula, ε2 i1 Indicates the reasonable degree of change of the printing feature of the i1th pixel; τ i1 represents the printing boundary weight coefficient of the i1th pixel; B represents the number of all pixels in the feature-related pixel sequence; Q b Indicates the degree of regional feature correlation of the b-th pixel; Q b+1 Indicates the degree of regional feature correlation of the b+1th pixel; || indicates taking the absolute value.
[0034] Preferably, the method of dividing all pixels according to the printing boundary degree of each pixel to obtain a number of ink spot defective pixels includes:
[0035] The pixel with the largest printing boundary degree in the feature-associated pixel sequence is recorded as the first target pixel, and each pixel before the first target pixel is recorded as an ink spot defect pixel.
[0036] Preferably, the method of adaptively adjusting the super pixel block size according to the printing boundary degree of each ink spot defect pixel point and performing printing anomaly detection includes the following specific methods:
[0037] A super pixel block size TL2 is preset, and the product of the printing boundary degree of each ink spot defect pixel and TL2 is used as the adaptive super pixel block size of each ink spot defect pixel; according to the adaptive super pixel block size of each ink spot defect pixel, the saliency value of each ink spot defect pixel is obtained by using the CA saliency detection algorithm;
[0038] A significance threshold TL3 is preset, and the image area occupied by ink spot defect pixels with a significance value greater than TL3 is recorded as an ink spot defect area; if any printed grayscale image contains an ink spot defect area, the printed grayscale image is recorded as a printed grayscale image with quality problems.
[0039] The beneficial effects of the technical solution of the present invention are as follows: the present invention obtains the printing boundary degree of each pixel point through the correlation of grayscale and gradient information between different pixel points and the dense distribution of corresponding positions, thereby screening out a number of ink spot defective pixels, and then adaptively adjusting the size of super pixel blocks for abnormality detection; firstly, the printing boundary degree of each pixel point is obtained according to the correlation of grayscale and gradient information between different pixel points and the dense distribution of corresponding positions, wherein the printing boundary degree is used to describe the degree to which the overall printing feature change trend of pixels before different pixels in the feature-associated pixel sequence tends to the feature change trend of ink spot defects, thereby reducing the interference of normal areas on pixel point identification of ink spot defect areas; the present invention obtains the printing boundary degree by analyzing the correlation of grayscale and gradient information between different pixel points and the dense distribution of corresponding positions, and adaptively adjusts the size of the super pixel block corresponding to each pixel point according to the printing boundary degree, thereby adaptively adjusting the significance of each pixel point, improving the accuracy of the detection result, and reducing the loss of normal pixels. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0041] Figure 1 The present invention is a structural block diagram of a printing anomaly detection and early warning system. DETAILED DESCRIPTION
[0042] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation, structure, features and effects of a printing abnormality detection and early warning system proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0043] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0044] The specific scheme of a printing anomaly detection and early warning system provided by the present invention is described in detail below with reference to the accompanying drawings.
[0045] See also Figure 1 , which shows a structural block diagram of a printing anomaly detection and early warning system provided by an embodiment of the present invention, the system includes the following modules:
[0046] The printed matter grayscale image acquisition module 101 acquires printed matter grayscale images of a plurality of printed matters.
[0047] It should be noted that the existing technology usually uses a CA (Context-Aware) saliency detection algorithm to detect ink spot defect areas in printed images. However, since some ink spots are similar to some printed text and patterns, some pixels belonging to the ink spot defect area will have saliency values similar to some normal pixels. Therefore, the ink spot defect area cannot be effectively detected, which reduces the accuracy of detecting the ink spot defect area and causes some normal pixels to be lost.
[0048] Specifically, it is first necessary to collect grayscale images of a number of printed products. The specific process is: use an industrial camera to shoot 50 printed product images; perform grayscale processing on each printed product image to obtain a number of printed product grayscale images.
[0049] It should be noted that the number of printed materials in this embodiment may be determined according to specific implementation conditions; grayscale processing is a well-known technology and will not be described in detail in this embodiment.
[0050] So far, the grayscale images of all printed products have been obtained through the above method.
[0051] The ink spot defect pixel recognition module 102 obtains the printing boundary degree of each pixel according to the correlation between the grayscale and gradient information of different pixels and the dense distribution of the corresponding positions; and divides all pixels according to the printing boundary degree of each pixel to obtain a number of ink spot defect pixels.
[0052] It should be noted that in the actual printing process, the color of the printed paper is usually a solid color background with bright tones, while the pattern or text printed on the paper is usually a solid color foreground with dark tones. When the printer is printing, due to the influence of factors such as the temperature, humidity and ink concentration of the surrounding environment, the printer may leak ink during printing, resulting in ink spots on the paper. The sizes, positions and shapes of these ink spots are randomly distributed, and each ink spot is a completely closed area with a low corresponding grayscale value; the image area occupied by the printed font is usually small and is also a completely closed area with a low corresponding grayscale value; resulting in ink spots of similar size to the printed font being unable to be effectively identified. However, since the size, position and shape of the printed font are relatively fixed and show a certain regularity, this embodiment can obtain the degree of printing defects of each pixel by analyzing the difference distribution of image features between the printed font and the ink spot for subsequent enhancement processing.
[0053] Preferably, in an embodiment of the present invention, the printing boundary degree of each pixel is obtained according to the correlation of grayscale and gradient information between different pixels and the dense distribution of corresponding positions, and the specific method includes:
[0054] Taking any printed grayscale image as an example, the sobel operator is used to obtain the gradient amplitude and gradient direction degree of all pixels in the printed grayscale image; a window size TL1 is preset, wherein this embodiment is described by taking TL1=7 as an example, and this embodiment is not specifically limited, wherein TL1 can be determined according to the specific implementation situation; any pixel point is recorded as a target pixel point, and a window area with a window size of TL1×TL1 centered on the target pixel point is used as a reference neighborhood area of the target pixel point; each pixel point in the reference neighborhood area of the target pixel point, except the target pixel point, is recorded as a reference pixel point of the target pixel point; taking the lth reference pixel point as an example, according to the difference in grayscale value, gradient amplitude, gradient direction degree and distance between the target pixel point and the lth reference pixel point, the printing feature correlation factor between the target pixel point and the lth reference pixel point is obtained. As an example, the printing feature correlation factor between the target pixel point and the lth reference pixel point can be calculated by the following formula:
[0055]
[0056] In the formula, q l represents the printing feature correlation factor between the target pixel and the lth reference pixel; H represents the gray value of the target pixel; H l A1 represents the gray value of the lth reference pixel; A2 represents the gradient amplitude of the target pixel; A3 represents the gray value of the lth reference pixel; A4 represents the gray value of the lth reference pixel; A5 represents the gray value of the lth reference pixel; A6 represents the gray value of the lth reference pixel; A7 represents the gray value of the lth reference pixel; A8 represents the gray value of the lth reference pixel; A9 represents the gray value of the lth reference pixel; A1 lrepresents the gradient magnitude of the lth reference pixel; θ represents the gradient direction degree of the target pixel; θ l Indicates the gradient direction degree of the lth reference pixel; d l represents the Euclidean distance between the target pixel and the lth reference pixel; || represents taking the absolute value. The acquisition of the Euclidean distance is a well-known technique and will not be described in detail in this embodiment.
[0057] It should be noted that if the printing feature correlation factor between the target pixel point and the lth reference pixel point is larger, it means that the target pixel point and the lth reference pixel point are more concentrated in distribution, the image information features between the pixels are more similar, which reflects that the printing features between the target pixel point and the lth reference pixel point are more similar.
[0058] Furthermore, the printing feature correlation factors of the target pixel and all the reference pixel points are linearly normalized, and each normalized printing correlation factor is recorded as the printing correlation degree; the standard deviation of the printing feature correlation degree between the target pixel and all the reference pixel points is used as the regional feature correlation degree of the target pixel point; the regional feature correlation degree of all the pixel points is obtained; the sequence of all the pixel points arranged in descending order according to the regional feature correlation degree is recorded as a feature correlation pixel point sequence; taking any two adjacent pixel points in the feature correlation pixel point sequence as an example, the difference in the feature correlation degree between the first pixel point and the second pixel point is recorded as the first difference, the Euclidean distance between the first pixel point and the second pixel point is recorded as the first distance, and the ratio of the first difference to the first distance is recorded as the printing boundary weight coefficient of the first pixel point.
[0059] It should be noted that this embodiment does not consider the printing boundary weight coefficient of the last pixel in the feature-associated pixel sequence.
[0060] Furthermore, according to the fluctuation difference of the regional feature association degree between adjacent pixels in the feature association pixel sequence and the printing boundary weight coefficient of the i1th pixel in the feature association pixel sequence, the printing boundary degree of the i1th pixel is obtained. As an example, the printing boundary degree of the i1th pixel can be calculated by the following formula:
[0061]
[0062] In the formula, μ i1 represents the printing boundary degree of the i1th pixel; Q1 represents the regional feature association degree of the first pixel in the feature association pixel sequence; Q i1 represents the regional feature correlation degree of the i1th pixel; Q2 represents the regional feature correlation degree of the last pixel in the feature-correlated pixel sequence; τ i1represents the printing boundary weight coefficient of the i1th pixel; B represents the number of all pixels in the feature-related pixel sequence; Q b Indicates the degree of regional feature correlation of the b-th pixel; Q b+1 It represents the degree of regional feature correlation of the b+1th pixel point; exp() represents an exponential function with a natural constant as the base. The embodiment adopts the exp(-x) model to present the inverse proportional relationship and normalization processing. x is the input of the model. The implementer can select the inverse proportional function and normalization function according to the actual situation; || represents taking the absolute value.
[0063] It should be noted that Indicates the rationality of the selection of the regional feature correlation degree of the i1-th pixel; Indicates the reasonable degree of change of the printing characteristics of the i1th pixel point; if the printing demarcation degree of the i1th pixel point is greater, it means that the overall printing characteristic change trend of the pixels before the i1th pixel point is more inclined to the characteristic change trend of the ink spot defect, reflecting that the i1th pixel point can be used as a demarcation point in the feature-associated pixel point sequence.
[0064] Preferably, in one embodiment of the present invention, all pixels are divided according to the printing boundary degree of each pixel to obtain a number of ink spot defective pixels, including the specific method of:
[0065] The pixel with the largest printing boundary degree in the feature-associated pixel sequence is recorded as the first target pixel, and each pixel before the first target pixel is recorded as an ink spot defect pixel.
[0066] So far, all the ink spot defective pixels are obtained through the above method.
[0067] The anomaly detection module 103 adaptively adjusts the size of the super pixel block according to the printing boundary degree of each ink spot defect pixel point, and performs printing anomaly detection.
[0068] Preferably, in one embodiment of the present invention, according to the printing boundary degree of each ink spot defect pixel point, the super pixel block size is adaptively adjusted, and printing anomaly detection is performed, including the specific method of:
[0069] A super pixel block size TL2 is preset, wherein this embodiment is described by taking TL2=5 as an example, and this embodiment is not specifically limited, wherein TL2 can be determined according to the specific implementation situation, and TL1≥TL2; the product of the printing boundary degree of each ink spot defect pixel and TL2 is used as the adaptive super pixel block size of each ink spot defect pixel; according to the adaptive super pixel block size of each ink spot defect pixel, the saliency value of each ink spot defect pixel is obtained by the CA saliency detection algorithm. The process of obtaining the saliency value of each pixel according to the super pixel block size of the pixel is a well-known content of the CA (Context-Aware) saliency detection algorithm, and will not be repeated in this embodiment.
[0070] Furthermore, a significance threshold TL3 is preset, wherein this embodiment is described by taking TL3=0.75 as an example, and this embodiment is not specifically limited, wherein TL3 may be determined according to the specific implementation situation; the image area occupied by ink spot defect pixels with a significance value greater than TL3 is recorded as an ink spot defect area; if the grayscale image of the printed product contains an ink spot defect area, the grayscale image of the printed product is recorded as a grayscale image of a printed product with quality problems.
[0071] At this point, this embodiment is completed.
[0072] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A printing anomaly detection and early warning system, It is characterized in that The system includes the following modules: A printed matter grayscale image acquisition module, used for acquiring printed matter grayscale images of a plurality of printed matters; The ink spot defect pixel recognition module is used to obtain the printing boundary degree of each pixel point according to the correlation between the grayscale and gradient information of different pixels and the dense distribution of the corresponding positions; all pixels are divided according to the printing boundary degree of each pixel point to obtain a number of ink spot defect pixels; An anomaly detection module is used to adaptively adjust the size of the super pixel block according to the printing boundary degree of each ink spot defect pixel point, and perform printing anomaly detection; The specific method of obtaining the printing boundary degree of each pixel point according to the correlation between the grayscale and gradient information of different pixel points and the dense distribution of the corresponding positions includes: For any printed grayscale image, use the Sobel operator to obtain the gradient amplitude and gradient direction of all pixels in the printed grayscale image; preset a window size , mark any pixel as the target pixel, take the target pixel as the center, and the window size is The window area is taken as the reference neighborhood area of the target pixel; each pixel point in the reference neighborhood area of the target pixel point, except the target pixel point, is recorded as the reference pixel point of the target pixel point; For reference pixel, based on the target pixel and the The difference in grayscale value, gradient amplitude and gradient direction between the first and second reference pixels is obtained. The same-direction correlation coefficient of the printing features of the reference pixels; According to the target pixel and the The printing feature same-direction correlation coefficient and corresponding distance of the reference pixel point are obtained. The printing feature correlation factor of the reference pixel; Linearly normalize the printing feature correlation factors between the target pixel and all reference pixels, and record each normalized printing correlation factor as the printing correlation degree; According to the distribution law of the printing feature correlation degree between the target pixel and all the reference pixels, the regional feature correlation degree of the target pixel is obtained; the regional feature correlation degree of each pixel is obtained; and the sequence of all the pixels arranged in descending order according to the regional feature correlation degree is recorded as the feature correlation pixel sequence; According to the degree of feature association and distribution aggregation between adjacent pixels in the feature-associated pixel sequence, a printing boundary weight coefficient of each pixel in the feature-associated pixel sequence is obtained; According to the feature association pixel sequence The difference in the degree of regional feature correlation between the pixel point and the two end pixels in the feature-correlated pixel point sequence is obtained. The reasonableness of the selection of the correlation degree of regional features of each pixel; According to the difference distribution of regional feature association degree between adjacent pixels in the feature association pixel sequence, and the The printing boundary weight coefficient of the pixel point is obtained Reasonable degree of change in printing characteristics of each pixel; The first The rationality of selecting the correlation degree of regional features of pixels is related to the The product of the reasonable degree of change of the printing characteristics of the pixels is taken as the first The degree of printing demarcation of pixels; The method according to the target pixel point and the The difference in grayscale value, gradient amplitude and gradient direction between the first and second reference pixels is obtained. The printing feature same-direction correlation coefficient of the reference pixel points includes the following specific methods: In the formula, Indicates the target pixel and the The same-direction correlation coefficient of the printing features of the reference pixels; Represents the gray value of the target pixel; Indicates Gray value of reference pixel; Represents the gradient amplitude of the target pixel; Indicates The gradient amplitude of the reference pixel; Indicates the gradient direction degree of the target pixel; Indicates The gradient direction degree of the reference pixel; Indicates taking the absolute value; The method according to the target pixel point and the The printing feature same-direction correlation coefficient and corresponding distance of the reference pixel point are obtained. The printing feature correlation factor of the reference pixel point includes the following specific methods: The target pixel is aligned with the The same-direction correlation coefficient of the printing characteristics of the reference pixel point and the target pixel point The ratio of the Euclidean distance between the target pixel and the The printing feature correlation factor of the reference pixel; The specific method of obtaining the regional feature correlation degree of the target pixel point according to the distribution law of the printing feature correlation degree between the target pixel point and all reference pixel points includes: The standard deviation of the correlation degree between the target pixel and all the reference pixels is used as the correlation degree of the regional features of the target pixel; The specific method of obtaining the printing boundary weight coefficient of each pixel in the feature-associated pixel sequence according to the feature association degree and distribution aggregation between adjacent pixels in the feature-associated pixel sequence is as follows: For any two adjacent pixels in the feature-associated pixel sequence, the difference in the feature association degree between the first pixel and the second pixel is recorded as the first difference, the Euclidean distance between the first pixel and the second pixel is recorded as the first distance, and the ratio of the first difference to the first distance is recorded as the printing boundary weight coefficient of the first pixel; The first pixel point on the feature-associated pixel sequence The difference in the degree of regional feature correlation between the pixel point and the two end pixels in the feature-correlated pixel point sequence is obtained. The reasonable degree of selection of the regional feature correlation degree of each pixel point includes the following specific methods: In the formula, Indicates The reasonableness of the selection of the correlation degree of regional features of each pixel; Indicates the regional feature correlation degree of the first pixel in the feature correlation pixel sequence; Indicates The degree of correlation between regional features of pixels; Indicates the regional feature correlation degree of the last pixel in the feature-correlated pixel sequence; represents an exponential function with a natural constant as base; Indicates taking the absolute value; The difference distribution of regional feature association degree between adjacent pixels in the feature association pixel sequence and the difference distribution of regional feature association degree between adjacent pixels in the feature association pixel sequence The printing boundary weight coefficient of the pixel point is obtained The reasonable degree of change of printing characteristics of each pixel point includes the following specific methods: In the formula, Indicates Reasonable degree of change in printing characteristics of each pixel; Indicates Printing boundary weight coefficient of each pixel; It represents the number of all pixels in the feature-associated pixel sequence; Indicates The degree of correlation between regional features of pixels; Indicates The degree of correlation between regional features of pixels; Indicates taking the absolute value; The specific method of dividing all pixels according to the printing boundary degree of each pixel to obtain a number of ink spot defective pixels includes: The pixel with the largest printing boundary degree in the feature-associated pixel sequence is recorded as the first target pixel, and each pixel before the first target pixel is recorded as an ink spot defect pixel; The method of adaptively adjusting the size of the super pixel block according to the printing boundary degree of each ink spot defect pixel and performing printing anomaly detection includes the following specific methods: Preset a superpixel block size , the printing boundary degree of each ink spot defect pixel is The product of is used as the adaptive superpixel block size of each ink spot defect pixel; according to the adaptive superpixel block size of each ink spot defect pixel, the saliency value of each ink spot defect pixel is obtained by using the CA saliency detection algorithm; Preset a significance threshold , the significant value is greater than The image area occupied by the ink spot defect pixel points is recorded as the ink spot defect area; if any printed grayscale image contains the ink spot defect area, the printed grayscale image is recorded as a printed grayscale image with quality problems.
