Print-through detection method and device based on image

Abstract

The embodiment of the invention provides an image-based print-through detection method and device, and the method comprises the steps: determining a to-be-detected image corresponding to a target document, carrying out the edge detection of the to-be-detected image, and obtaining an edge feature image corresponding to the to-be-detected image; determining a feature region image based on the edge feature image, and calculating a first histogram feature corresponding to the to-be-detected image and a second histogram feature corresponding to the feature region image; determining an image association degree between the to-be-detected image and the feature region image according to the first histogram feature and the second histogram feature, and determining content attribute information corresponding to the feature region image based on the image association degree; and generating a print-through detection result corresponding to the to-be-detected image according to the content attribute information. The image correlation degree is determined through the histogram features so as to determine whether the image in the feature region has a print-through phenomenon, and the detection precision and efficiency are improved.

Description

TECHNICAL FIELD

[0001] Embodiments of the present specification relate to the technical field of image processing, in particular to a kind of image-based print-through detection method and device. BACKGROUND

[0002] The electronicization of paper documents is an important means of preserving documents, which scans paper documents into text images that are easier to store and copy on electronic devices through scanning. Paper documents include printed paper documents and handwritten paper documents, and the content of both types of paper documents may appear print-through phenomenon. Therefore, the front and back of the content on the paper document needs to be distinguished. However, the prior art cannot clearly distinguish the front content and the print-through content, which leads to the subsequent inability to effectively remove the print-through content. At the same time, manual correction using image editing software is often used in the removal process, which is not only complicated but also time-consuming and labor-intensive. Therefore, how to effectively distinguish the effective content or print-through content on the paper document to facilitate the subsequent removal of the print-through content is a problem that needs to be solved at present. SUMMARY

[0003] Therefore, the embodiments of the present specification provide an image-based print-through detection method. One or more embodiments of the present specification also relate to an image-based print-through detection device, a computing device, a computer-readable storage medium and a computer program product to solve the technical defects in the prior art.

[0004] According to a first aspect of the embodiments of the present specification, an image-based print-through detection method is provided, comprising: determining a target document corresponding to a to-be-detected image, performing edge detection on the to-be-detected image to obtain an edge feature image corresponding to the to-be-detected image; determining a feature region image based on the edge feature image, and calculating a first histogram feature corresponding to the to-be-detected image and a second histogram feature corresponding to the feature region image; determining an image correlation degree between the to-be-detected image and the feature region image according to the first histogram feature and the second histogram feature, and determining content attribute information corresponding to the feature region image based on the image correlation degree; generating a print-through detection result corresponding to the to-be-detected image according to the content attribute information.

[0005] According to a second aspect of the embodiments of the present specification, an image-based print-through detection device is provided, comprising: The detection module is configured to determine a target document corresponding to a to-be-detected image, perform edge detection on the to-be-detected image to obtain an edge feature image corresponding to the to-be-detected image; The computing module is configured to determine a feature region image based on the edge feature image, and calculate a first histogram feature corresponding to the to-be-detected image and a second histogram feature corresponding to the feature region image. The determining module is configured to determine an image correlation degree between the to-be-detected image and the feature region image according to the first histogram feature and the second histogram feature, and determine content attribute information corresponding to the feature region image based on the image correlation degree. The generating module is configured to generate a print-through detection result corresponding to the to-be-detected image according to the content attribute information.

[0006] According to a third aspect of an embodiment of the present specification, a computing device is provided, comprising: a memory and a processor; The memory is configured to store computer executable instructions, and the processor is configured to execute the computer executable instructions, and the computer executable instructions, when executed by the processor, implement the steps of the image-based print-through detection method.

[0007] According to a fourth aspect of an embodiment of the present specification, a computer readable storage medium is provided, which stores computer executable instructions, and the instructions, when executed by a processor, implement the steps of the image-based print-through detection method.

[0008] According to a fifth aspect of an embodiment of the present specification, a computer program product is provided, comprising a computer program or instructions, and the computer program or instructions, when executed by a processor, implement the steps of the image-based print-through detection method.

[0009] One embodiment of the present specification realizes edge detection on a to-be-detected image corresponding to a target document to obtain a corresponding edge feature image, determines a feature region image based on the edge feature image, calculates a first histogram feature corresponding to the to-be-detected image and a second histogram feature corresponding to the feature region image, and determines an image correlation degree between the to-be-detected image and the feature region image using the first histogram feature and the second histogram feature. Thus, the content attribute information corresponding to the feature region image is determined using the image correlation degree, and a print-through detection result of the to-be-detected image is generated according to the content attribute information. The histogram feature is used to determine the image correlation degree to determine whether the feature region image has a print-through phenomenon, thereby improving the detection accuracy and efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0010] Figure 1 is a flowchart of an image-based print-through detection method provided by one embodiment of the present specification; Figure 2 is a process flowchart of an image-based print-through detection method provided by one embodiment of the present specification; Figure 3 is a structural schematic diagram of an image-based see-through detection device provided by an embodiment of the present specification; Figure 4 is a structural block diagram of a computing device provided by an embodiment of the present specification. DETAILED DESCRIPTION

[0011] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present specification. However, the present specification can be practiced without the specific details, other than in the examples described herein, and it is understood that the scope of the present specification is not limited to the details below.

[0012] The terminology used in one or more embodiments of the present specification is for the purpose of describing particular embodiments only and is not intended to be limiting of one or more embodiments of the present specification. As used in one or more embodiments of the present specification and the accompanying claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in one or more embodiments of the present specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0013] It will be understood that, although the terms first, second, etc. can be used herein to describe various information, these terms are not intended to denote a temporal or chronological order. Rather, these terms are used solely to distinguish one from another only. For example, without departing from the scope of one or more embodiments of the present specification, first can be termed second, and similarly, second can be termed first. The word "if' as used herein can be interpreted as meaning "when" or "upon" or "in response to determining" depending on the context.

[0014] In addition, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in one or more embodiments of the present specification are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation portal for user to choose authorization or refusal.

[0015] First, the terms involved in one or more embodiments of the present specification are explained.

[0016] Image bleed: Image bleed refers to the phenomenon that when scanning or photocopying double-sided printed paper, due to the thinness of the paper, the text or patterns on the back side will affect the content on the front side. This phenomenon will cause the content on the front and back sides to interweave in the scanned image, reducing the clarity and readability of the image, and bringing challenges to document digitization, content recognition, etc.

[0017] Edge detection: Edge detection is an image processing technique used to identify areas with significant brightness changes in an image to determine the boundaries of objects. Common methods include Sobel operator (Sobel operator is an important algorithm in image processing, mainly used for edge detection. It identifies edges in an image by calculating the gradient of each pixel point in the image), Canny operator (Canny operator aims to achieve efficient edge detection through a multi-stage algorithm, and minimize error rate as much as possible while ensuring the detected edges have good positioning accuracy), etc. It can help extract key feature information of the image.

[0018] Content back-through detection in single-sided printing scenarios: When the paper is single-sided printed, it is difficult to accurately determine whether the content in the scanned image is the front printed text pattern or the text pattern from the back due to the thinness of the paper. The existing technology has obvious shortcomings in distinguishing between the two cases, and the analysis of the light transmission characteristics of paper materials and the features of printed content is not deep enough. This makes it easy to misjudge in scenarios such as document digitization and content recognition, and cannot effectively guarantee the accuracy and purity of information, thereby affecting the further processing and utilization of single-sided printed paper.

[0019] Content differentiation in double-sided printing scenarios: For double-sided printed paper, the content on the front and back sides of the scanned image interweaves, making it difficult to clearly distinguish which parts are the front normal printed content and which parts are the back through content. Existing double-sided printing content processing technology has difficulty in accurately distinguishing between the two types of content, and often can only perform some simple overall processing, without effectively separating at the pixel level or feature level, resulting in the inability to process the front and back through content separately in the process of document editing and content proofreading, which seriously affects work efficiency and quality.

[0020] Noise removal in double-sided printing scenarios: In the scanned image of double-sided printed paper, the back through content will interfere with the normal printed content on the front, reducing the clarity and readability of the image. However, current common image processing techniques are not effective in removing this back content noise caused by paper back-through, or may damage important details of the front normal printed content during the removal process, failing to achieve the desired denoising effect, making it impossible to effectively improve the appearance and quality of the front printed content, and unable to meet the requirements of high-quality document image processing.

[0021] Traditional image processing algorithms, such as blurring and denoising, primarily target image noise and background noise. These algorithms are limited in their effectiveness when processing text or patterns that show through from the front. This is because backside see-through isn't just a noise issue; it's the mutual interference of text or patterns on both sides. In this case, simple blurring or denoising cannot effectively distinguish and eliminate specific interfering elements, resulting in a noticeable backside see-through effect in the processed image, which can affect the clarity of the content. While duplex scanning compensation algorithms can recognize frontside see-through text, they primarily rely on the optical properties of the image and may not completely eliminate see-through interference. This is especially true when the text or pattern has complex colors and shapes, significantly reducing the accuracy and effectiveness of the algorithm.

[0022] Manual correction using image editing software, while capable of fine-tuning specific situations, is costly, inefficient, and unsuitable for large-scale scanning tasks. Each image requires manual processing, making it difficult to scale and particularly inconvenient for large-scale printing and scanning. Manual adjustments during printing, such as selecting paper and adjusting equipment pressure, rely on the operator's skill and experience, resulting in inconsistent quality and slow processing speeds. This method is insufficient for large-scale document or book scanning. Therefore, manual image processing methods face significant limitations in practical application. Furthermore, they cannot process content that has already been printed.

[0023] In response to the above problems, this specification provides an image-based print-through detection method. This specification also involves an image-based print-through detection device, a computing device, a computer-readable storage medium, and a computer program product, which are described in detail one by one in the following embodiments.

[0024] See also Figure 1 , Figure 1 A flowchart of an image-based print-through detection method provided according to an embodiment of the present specification is shown, which specifically includes the following steps.

[0025] Step 102: Determine an image to be detected corresponding to the target document, perform edge detection on the image to be detected, and obtain an edge feature image corresponding to the image to be detected.

[0026] The target document can be understood as a target paper document, and the target document is printed with corresponding document data. In the printing process, due to the thinness of the paper and the heaviness of the ink, the paper may be transparently printed. In order to facilitate subsequent normal use of the target document for image scanning or browsing of the electronic document by the user, it is necessary to detect and identify the paper transparent printing of the target document, and in the case of paper transparent printing, the transparent printing area and the normal area need to be distinguished, so as to remove the transparent printing area. The image-based transparent printing detection method provided in the specification aims to achieve the above-mentioned purpose.

[0027] In actual application, first, the target document can be image collected such as scanning, shooting and other collection means to obtain the corresponding target document detection image. The edge feature image in the detection image is identified by edge detection. The edge feature image is an image containing the edge features in the detection image. The edge feature is the area where the pixel gray value in the image changes sharply, which usually corresponds to the object boundary or contour in the image. According to the edge feature, the related content in the detection image can be identified.

[0028] Further, the target document corresponding to the detection image is determined, including: determining the initial detection image corresponding to the target document; and performing noise reduction processing on the initial detection image to obtain the target document corresponding to the detection image.

[0029] The initial detection image can be understood as an image obtained by image collection of the target document. In order to reduce the noise in the image, the initial detection image can be processed by noise reduction to obtain the noise-reduced detection image.

[0030] In actual application, the noise reduction processing can be understood as processing the initial detection image by using a noise reduction algorithm. In the embodiment of the specification, Gaussian blur algorithm is used for noise reduction processing, which reduces noise and details by smoothing the initial detection image. The Gaussian blur algorithm is realized by processing the image and the Gaussian kernel through convolution operation. The weight of the Gaussian kernel follows the Gaussian distribution, and each element of the Gaussian kernel is calculated according to its distance from the center. The formula of Gaussian distribution is as follows:

[0031] Where G(x, y) is the value of Gaussian distribution, x and y are the horizontal and vertical coordinates of the pixel distance from the kernel center, and σ is the standard deviation, which controls the width of the Gaussian distribution.

[0032] In practice, a Gaussian kernel is a matrix that defines how an image is weighted within a local neighborhood. In image processing, the new value of each pixel is determined by the weighted average of the pixel itself and its surrounding pixels, with the weights determined by the values ​​in the Gaussian kernel. The Gaussian kernel used in the Gaussian blur algorithm is based on a two-dimensional Gaussian function. Its characteristic is that the weight is highest at the center and decreases symmetrically around the edges, resulting in a "soft" blur effect.

[0033] In one embodiment of this specification, a 5×5 matrix can be used as a Gaussian kernel. This means that when blurring, each pixel is weighted against the 25 pixels surrounding it in 5 rows and 5 columns. Compared to smaller kernels (such as 3×3), a 5×5 kernel offers stronger denoising and smoothing capabilities, making it suitable for initially removing high-frequency noise in an image. At the same time, compared to larger kernels (such as 7×7), it minimizes detail loss, maintaining a good balance between processing efficiency and quality.

[0034] It should be noted that in image processing, images are typically converted to grayscale before blurring to simplify calculations or unify color spaces. However, subsequent methods require statistical histograms and correlation analysis, and the histogram of a color image can contain information from multiple dimensions (such as the distribution of the three RGB channels). If the image is converted to grayscale beforehand, some color features will be lost, affecting the accuracy and discriminative power of subsequent histogram features. Back-transparent content may appear differently in different color channels, and preserving color information helps improve the algorithm's ability to distinguish between front and back content. Especially under certain paper materials or printing ink conditions, the contrast between color channels can be a key clue in identifying back-transparent content.

[0035] Based on this, a 5×5 Gaussian kernel is used for blurring, which effectively removes image noise while maintaining sufficient image details. The image is not converted to grayscale throughout the entire process in order to better support subsequent multi-channel histogram statistics and correlation analysis, thereby improving the recognition accuracy and robustness of the front and back content in the image.

[0036] Furthermore, edge detection is performed on the image to be detected to obtain an edge feature image corresponding to the image to be detected, including: performing edge detection on the image to be detected to determine the image gradient of the image to be detected; determining image edge information in the image to be detected based on the image gradient, and generating an edge feature image corresponding to the image to be detected based on the image edge information.

[0037] Among them, after the Gaussian blur processing, the image to be detected can be obtained, and then edge detection can be performed on the image to be detected to obtain an edge feature image corresponding to the image to be detected.

[0038] In practical applications, applying an edge detection algorithm (such as the Canny edge detection algorithm) to the image to be detected, the first step of the process is to determine the gradient of the image. By using convolution operations, the partial derivatives of the image in the x and y directions are calculated respectively, resulting in the gradient magnitude and direction at each pixel point. The formula for calculating the gradient is as follows:

[0039] where I is the gray value of the image, Gx and Gy are the gradients of the image in the x and y directions, and G is the gradient magnitude.

[0040] After obtaining the image gradient, in order to refine the edges and reduce the influence of false edges, the non-maximum suppression technique is used. Non-maximum suppression (NMS) is a technique used to refine edges, especially in image processing and computer vision tasks for edge detection algorithms (such as the Canny operator). The main purpose of NMS is to remove redundant response points in the edge detection process and retain the pixels that are most likely to represent the true edges, resulting in more accurate and clear edges. This process retains the points of local maximum gradient magnitude and suppresses other points that are not maximum, ensuring that only the pixels that are most likely to belong to the edges are retained. Next, it is necessary to further determine which edges are the true edges, by setting two thresholds (for example, 50 and 150), the high threshold is used to identify clear strong edges, and the low threshold is used to connect weak edges. Only those pixels that exceed the high threshold or are between the high and low thresholds and are connected to strong edges will be considered valid edges.

[0041] In specific implementation, a double threshold is applied by the edge detection algorithm to detect and connect edges. The low threshold is used to connect edges, and the high threshold is used to detect strong edges. This process allows the edge detection algorithm not only to detect obvious edges but also to connect weaker edges, resulting in a coherent edge image.

[0042] According to the above processed image edge information, the final edge feature image is generated. In the edge feature image, only the pixels identified as edges are retained, and the rest are usually set to a background color (such as black), forming a binary image that clearly displays the edge structure of the image.

[0043] In an embodiment of the present disclosure, edge detection is performed on the image to be detected, and a preset edge detection algorithm is used to calculate the gradient of the image in the x-axis and y-axis directions, and then the gradient amplitude and direction of each pixel position are obtained. For each pixel, if its gradient amplitude is locally maximum in its gradient direction, its value is retained; otherwise, it is set to 0 (i.e., not an edge). Two high and low thresholds are set, all pixels with a gradient amplitude greater than the high threshold are marked as strong edge points, all pixels with a gradient amplitude less than the low threshold are marked as non-edge points, and pixels between the two thresholds are determined whether to be retained according to whether they are connected to strong edge points. Finally, an edge feature image is generated based on the processing results described above. In this image, all pixels marked as edges are retained and can be highlighted, and non-edge pixels are removed or set to a background color.

[0044] Based on this, the key edge information can be extracted from the original image through the above steps to form an edge feature image clearly showing the edge structure of the image, thereby providing a basis for further content analysis.

[0045] Step 104: determining a feature region image based on the edge feature image, and calculating a first histogram feature corresponding to the image to be detected and a second histogram feature corresponding to the feature region image.

[0046] The feature region image can be understood as a region image determined from the edge feature image for detection. Since the target document can be single-sided printed or double-sided printed, it is necessary to determine whether the image content in the image to be detected is front content or back content for single-sided printing, and to determine which is front content and which is back content for double-sided printing. Therefore, the feature region image used in detection is different for different printing conditions. In the case of single-sided printing, the feature region image is the entire edge feature image, and in the case of double-sided printing, the feature region image needs to be obtained by splitting the edge feature from the edge feature image. Subsequently, the first histogram feature corresponding to the image to be detected and the second histogram feature corresponding to the feature region image can be calculated to determine whether the feature region image is front content or back content.

[0047] Further, determining a feature region image based on the edge feature image includes: determining an image detection strategy corresponding to the image to be detected; in the case of a single-sided detection strategy, using the edge feature image as the feature region image; and in the case of a double-sided detection strategy, splitting the edge feature image to obtain a plurality of edge feature sub-images as the feature region image.

[0048] The single-side detection strategy can be understood as a strategy for performing see-through content detection on the to-be-detected image of the target document in the case of single-side printing of the target document. For a single-side printed document, the main challenge is to determine whether the content in the to-be-detected image is the content of the front side or the text pattern from the back side. Because all the extracted edge information comes from the same side in single-side printing, there is no need for further splitting, and therefore the entire edge feature image is regarded as the feature region image in the case of the single-side detection strategy.

[0049] The double-side detection strategy can be understood as a strategy for performing see-through content detection on the to-be-detected image of the target document and distinguishing the see-through content and the front-side content in the case of double-side printing of the target document. For the case of double-side printing, the problem becomes more complex because it is necessary to accurately distinguish which part belongs to the front-side content and which part is the content from the back side. Therefore, in the case of double-side detection, the edge feature image needs to be split to obtain multiple edge feature sub-images as the feature region images. That is, this step involves morphological operations (such as dilation, erosion), gray level analysis, and application of adaptive threshold algorithms to segment the image into multiple edge feature sub-images corresponding to different clusters according to the edge distribution and intensity difference.

[0050] In actual applications, in the case of single-side printing, the entire edge feature image can be directly used for histogram correlation analysis to complete the detection task. In the case of double-side printing, the edge feature image needs to be split through a series of image processing techniques (including but not limited to edge detection, morphological operation, gray level analysis, etc.), and then the histogram feature calculation and analysis are performed on each sub-image to accurately distinguish the front-side and back-side content.

[0051] Therefore, by using different detection strategies for different printing conditions, the accuracy of detection is improved, and the back-through problem caused by thin paper can be effectively addressed, which is suitable for various application scenarios such as document digitization and content recognition.

[0052] Further, the first histogram feature corresponding to the to-be-detected image and the second histogram feature corresponding to the feature region image are calculated, including: in the case of the image detection strategy being a single-side detection strategy, determining first gray information corresponding to the to-be-detected image, calculating the first histogram feature corresponding to the to-be-detected image according to the first gray information; determining second gray information corresponding to the edge feature image, and calculating the second histogram feature corresponding to the feature region image according to the second gray information.

[0053] When using a single-sided detection strategy, it is necessary to calculate the histogram features of the image to be detected and the feature region image set, i.e., the entire edge feature image. Histograms are an important tool in image processing, used to display the distribution of pixel values ​​in an image. In image analysis, histograms can help understand image characteristics such as brightness, contrast, and color distribution.

[0054] In practical applications, if the image to be detected is a color image, it is first converted into a grayscale image. This step is necessary because histogram analysis is usually performed based on grayscale. Traverse each pixel in the grayscale image and record the frequency of occurrence of each grayscale level. For an 8-bit grayscale image, the grayscale level range is 0 to 255. Based on the traversal results, the grayscale information of each pixel in the image to be detected is determined, thereby obtaining the first grayscale information corresponding to the image to be detected. Based on the first grayscale information, the first histogram feature corresponding to the image to be detected can be calculated. The specific calculation formula of the histogram feature is as follows:

[0055] Where H(i) is the histogram value of gray level i, M and N are the width and height of the image, and I jk is the pixel value of the image at position (j, k), I (I jk =i) is an indicator function that takes the value 1 when the pixel value is equal to gray level i and takes the value 0 otherwise.

[0056] For the single-sided detection strategy, the entire edge feature image is directly used as the feature region image. The edge feature image is obtained through the edge detection algorithm mentioned above, which highlights the edge part of the image. Although the edge feature image is usually binarized (the edge is white or other significant colors, and the background is black), grayscale statistics can still be performed on it. This is because even the binarized image may contain some gray pixels of varying degrees generated during the edge detection process. Accordingly, each pixel point of the entire edge feature image is traversed, the frequency of occurrence of each gray level is recorded, and the second grayscale information corresponding to the edge feature image is determined based on the traversal result. Based on the second grayscale information, the second histogram feature corresponding to the entire edge feature image of the feature region image set can be calculated.

[0057] Based on this, we statistically analyzed the grayscale information of the original image to be inspected and its edge feature image, and calculated their respective histogram features. These histogram features provide basic data for subsequent correlation analysis of histogram features, helping to detect whether the image to be inspected has print-through. In the case of single-sided printing, by comparing the correlation between the two histograms, we can effectively determine the presence of back-through and its impact.

[0058] Further, the edge feature image is split to obtain a plurality of edge feature sub-images as feature region images, including: extracting edge point features in the edge feature image; splitting the edge feature image based on the edge point features to obtain a plurality of edge feature sub-images as feature region images.

[0059] In processing double-sided printed documents, the edge feature image needs to be further split to distinguish the front and back contents. This process involves extracting edge point features from the edge feature image, which can be understood as features about edge points extracted from the edge feature region, and segmenting the image based on these edge point features to obtain a plurality of edge feature sub-images as feature region images.

[0060] In practical applications, morphological operations such as dilation and erosion are used to enhance edge features, making the boundaries between different content regions clearer, which helps subsequent segmentation operations. Although the edge feature image is binary, in some cases, some grayscale information may be retained. By analyzing this grayscale information, the distribution and intensity of the edge can be better understood. Adaptive thresholding algorithms such as local mean value calculation are applied to dynamically adjust the threshold according to the overall brightness changes of the image to adapt to different lighting conditions and background changes, improving the accuracy of segmentation.

[0061] In specific implementation, morphological operations can help us segment based on the morphological features of the edge, and combine grayscale level analysis to classify the edge. Using dilation and erosion operations in morphological processing to enhance features of specific morphology, morphological processing includes dilation, erosion, etc., and subsequent grayscale level analysis can be combined to split the edge, thereby extracting each edge point feature. According to the overall distribution of the image, an adaptive thresholding algorithm is applied to calculate a dynamic threshold to avoid presetting to adapt to a wider range of scenarios and enhance the robustness of the method. Local mean value is used for calculation, and the formula is as follows:

[0062] where T local (x, y) is the average value of the local image block W x H, and I(i, j) is the grayscale value of the image at position (i, j).

[0063] After the binarization segmentation process, we will split each feature component according to the morphological, binary split, and determine the edge point features according to the split results. (On the basis of morphology, part of the dark area, part of the light area, and so on), specifically, the pixel points with similar edge intensity or direction are gathered together, and the commonly used methods include K-means clustering, etc. Identify and separate each connected domain in the image. Each connected domain represents a possible independent content area (may be part of the front content or part of the back transparent printing). According to the clustering results or the results of the connected domain analysis, determine the multiple edge feature sub-images corresponding to each edge point feature. Each edge feature sub-image should contain only one type of content (front or back) as much as possible.

[0064] Based on this, through the above steps, multiple edge feature sub-images can be extracted from the edge feature image, each sub-image represents a potential content area. This method not only effectively helps to distinguish the front and back content, but also provides more accurate basic data for subsequent histogram feature calculation and content recognition. This method is especially suitable for processing double-sided printed documents, and can significantly improve the accuracy and efficiency of recognition.

[0065] Step 106: Determine the image correlation degree between the to-be-detected image and the feature region image according to the first histogram feature and the second histogram feature, and determine the content attribute information corresponding to the feature region image based on the image correlation degree.

[0066] Among them, the image correlation degree between the to-be-detected image and the feature region image can be determined according to the first histogram feature and the second histogram feature. The image correlation degree can be understood as the correlation between two images. According to the image correlation degree, the content attribute information corresponding to the feature region image can be determined. The content attribute information includes front attribute or back attribute, that is, the content attribute information is used to reflect whether the image content in the feature region image is front printed content or back transparent content.

[0067] In actual application, the image correlation degree between the to-be-detected image and the feature region image can be determined according to the first histogram feature and the second histogram feature. The image correlation degree is used to quantify the similarity between two images. The calculation formula of the image correlation degree is as follows:

[0068] Among them, H1 and H2 are the first histogram feature and the second histogram feature respectively, μ H1 and μ H2The mean values of the first and second histogram features, respectively, are calculated to evaluate the similarity between the two histogram features as the image correlation degree between the to-be-detected image and the feature region image. The image correlation degree is between -1 and 1. When the image correlation degree is 1, it means that the two histogram features are the same, 0 means that the two histogram features are irrelevant, and -1 means that the two histogram features are opposite.

[0069] Specifically, when the two histogram features are determined to be the same according to the image correlation degree, it means that the to-be-detected image and the feature region image are the same, indicating that the content in the feature region image is a positive content. On the contrary, when the two histogram features are determined to be opposite according to the image correlation degree, it means that the to-be-detected image and the feature region image are opposite, indicating that the content in the feature region image is a negative content. Therefore, the content attribute information of the feature region image, i.e., whether it belongs to a positive content or a negative content, can be determined according to the calculated image correlation degree.

[0070] Further, the first histogram feature corresponding to the to-be-detected image and the second histogram feature corresponding to the feature region image are calculated, including: determining a plurality of to-be-detected sub-images corresponding to the plurality of edge feature sub-images in the to-be-detected image; determining first gray sub-information corresponding to the plurality of to-be-detected sub-images, and calculating a plurality of first histogram sub-features corresponding to the plurality of to-be-detected sub-images according to the first gray sub-information; determining second gray sub-information corresponding to the plurality of edge feature sub-images, and calculating a plurality of second histogram sub-features corresponding to the plurality of edge feature sub-images according to the second gray sub-information.

[0071] In the case of double-sided printing detection, the feature region image includes a plurality of edge feature sub-images, and the detection of positive and negative contents of each edge feature sub-image is required, i.e., it is required to determine whether each edge feature sub-image contains a negative content. By determining the first gray sub-information corresponding to each to-be-detected sub-image, the first histogram sub-feature corresponding to each to-be-detected sub-image is calculated according to the first gray sub-information. Correspondingly, by determining the second gray sub-information corresponding to each edge feature sub-image, a plurality of second histogram sub-features corresponding to the plurality of edge feature sub-images are calculated according to the second gray sub-information.

[0072] In practical applications, firstly, a plurality of to-be-detected sub-images corresponding to the plurality of edge feature sub-images in the to-be-detected image need to be determined, which can be specifically selecting to-be-detected sub-images with the same region position information in the to-be-detected image according to the region position information of the edge feature sub-images in the edge feature map. The first histogram sub-features and the second histogram sub-features are obtained by respectively calculating the histogram sub-features corresponding to the edge feature sub-images and the to-be-detected sub-images. Subsequently, the image sub-correlation degrees between each group of edge feature sub-images and to-be-detected sub-images can be determined according to the first histogram sub-features and the second histogram sub-features, so as to realize the differentiation of the image content in the edge feature sub-images.

[0073] Based on this, by determining the corresponding to-be-detected sub-images in the to-be-detected image according to the edge feature sub-images, the first histogram sub-features and the second histogram sub-features corresponding to the edge feature sub-images and the to-be-detected sub-images are calculated, which facilitates the subsequent determination of the content attribute information of the edge feature sub-images based on the first histogram sub-features and the second histogram sub-features.

[0074] Further, the image correlation degree between the to-be-detected image and the feature region image is determined according to the first histogram features and the second histogram features, including: determining the image sub-correlation degrees between the plurality of to-be-detected sub-images and the plurality of edge feature sub-images according to the first histogram sub-features and the second histogram sub-features; determining the image correlation degree between the to-be-detected image and the feature region image according to the image sub-correlation degrees.

[0075] Wherein, in order to further refine the process of determining the image correlation degree between the to-be-detected image and the feature region image according to the first histogram features and the second histogram features, the concept of "image sub-correlation degree" can be introduced. This method allows us to more accurately evaluate the similarity between each to-be-detected sub-image and the corresponding edge feature sub-image, and to comprehensively judge the overall correlation degree between the entire to-be-detected image and the feature region image based on these local similarities.

[0076] In practical applications, since there is a corresponding relationship between the edge feature sub-image and the to-be-detected sub-image, the image sub-correlation degree between the edge feature sub-image and the to-be-detected sub-image can be calculated according to the first histogram sub-features and the second histogram sub-features corresponding to the two, and based on the image sub-correlation degree, the image content in the edge feature sub-image can be correctly analyzed as positive content or negative content.

[0077] Therefore, the image correlation degree between the to-be-detected image and the feature region image can be effectively determined according to the first histogram feature and the second histogram feature through the above steps, and the front-side printing content and the back-side transparent printing content can be accurately distinguished based on the correlation degree. This method not only improves the processing accuracy, but also can adapt to different complexity of document image processing tasks.

[0078] Step 108: generating a transparent printing detection result corresponding to the to-be-detected image according to the content attribute information.

[0079] The content attribute information includes the front-back detection information of the edge feature image. In the case of single-side printing detection, it can be judged according to the content attribute information that the image content in the to-be-detected image is front-side content or back-side content, that is, it is determined whether the to-be-detected image is placed front-side or back-side, so that the transparent printing detection result can be generated as no transparent printing content or transparent printing content in the to-be-detected image. In the case of double-side printing detection, it can be judged according to the content attribute information that which to-be-detected sub-image in the to-be-detected image has front-side content and which to-be-detected sub-image has back-side content. Therefore, the transparent printing detection result can not only include whether the to-be-detected image contains transparent printing content, but also include which part of the image content is transparent printing content.

[0080] Further, after generating the transparent printing detection result corresponding to the to-be-detected image according to the content attribute information, the method further comprises: determining a target edge feature sub-image and a target to-be-detected sub-image corresponding to the target edge feature sub-image in the plurality of edge feature sub-images according to the transparent printing detection result; performing image optimization on the target to-be-detected sub-image to obtain an optimized target to-be-detected sub-image; and generating a target image according to the optimized target to-be-detected sub-image.

[0081] After obtaining the transparent printing detection result of the to-be-detected image, the transparent printing content in the to-be-detected image can be repaired, so that the to-be-detected image can be normally used. Therefore, the target edge feature sub-image can be determined in the plurality of edge sub-images according to the transparent printing detection result, the target edge feature sub-image is the edge sub-image belonging to the transparent printing content, and the target to-be-detected sub-image corresponding to the target edge feature sub-image can be understood as the to-be-detected sub-image in the to-be-detected image according to the region position of the target edge feature sub-image. After performing image optimization on the target to-be-detected sub-image, an optimized target to-be-detected sub-image can be obtained, and a target image can be generated according to the optimized target to-be-detected sub-image, the target image being an image without transparent printing content.

[0082] In practical applications, image optimization on the target sub-image to be detected can include image pixel filling, interpolation optimization, etc. In specific implementation, after the region position of the target sub-image to be detected in the image to be detected is determined, the region position can be appropriately enlarged to ensure the consistent and aesthetic nature of subsequent processing of the surrounding image. The erased region is effectively filled through image pixel filling, interpolation optimization, etc. so as to be consistent with the surrounding pixel values and ensure the natural and continuous visual effect. Specifically, the process of filling the erased region can be performed by using the average value of the neighboring pixels. The pixels in the region to be filled are replaced by the average value of the neighboring pixels, so that the filled region is smoothly transitioned with the surrounding pixels. After the average value filling is completed, interpolation optimization technology can be further used to solve the consistency problem of the surrounding. The interpolation technology allows the target region to be filled by calculating the values of the neighboring pixels, so as to be consistent with the surrounding environment. The bilinear interpolation algorithm is adopted to fill the target region by calculating the weighted average value of the four neighboring pixels at the position to be filled, which can effectively make the gray scale of the filled image gradually transition and maintain consistency. This processing method can ensure that the processed image has good visual effect and overall consistency even in a complex scene. Accordingly, the overall smoothing processing can be further performed on the external region, such as the blur processing by the median filter, so that the removal of the transparent content does not affect the visual effect and overall consistency of the overall image.

[0083] The image-based see-through detection method provided in the specification includes determining a target document corresponding to a to-be-detected image, performing edge detection on the to-be-detected image to obtain an edge feature image corresponding to the to-be-detected image, determining a feature region image based on the edge feature image, calculating a first histogram feature corresponding to the to-be-detected image and a second histogram feature corresponding to the feature region image, determining an image correlation degree between the to-be-detected image and the feature region image according to the first histogram feature and the second histogram feature, determining content attribute information corresponding to the feature region image based on the image correlation degree, and generating a see-through detection result corresponding to the to-be-detected image according to the content attribute information. The edge feature image corresponding to the target document is obtained by performing edge detection on the to-be-detected image. The feature region image is determined based on the edge feature image. The first histogram feature corresponding to the to-be-detected image and the second histogram feature corresponding to the feature region image are calculated. The image correlation degree between the to-be-detected image and the feature region image is determined by using the first histogram feature and the second histogram feature. The content attribute information corresponding to the feature region image is determined by using the image correlation degree. The see-through detection result of the to-be-detected image is generated according to the content attribute information. The image correlation degree is determined by using the histogram feature to determine whether the feature region image has a see-through phenomenon, thereby improving the detection accuracy and efficiency.

[0084] The following description is combined with the accompanying drawings Figure 2With the application of the image-based see-through printing detection method provided in the specification to single-side printing detection as an example, the image-based see-through printing detection method is further described. In this regard, Figure 2 A processing process flowchart of an image-based see-through printing detection method provided in an embodiment of the specification is shown, which specifically includes the following steps.

[0085] Step 202: Determine an initial to-be-detected image corresponding to a target document, perform noise reduction processing on the initial to-be-detected image, and obtain a to-be-detected image corresponding to the target document.

[0086] Step 204: Perform edge detection on the to-be-detected image, determine an image gradient of the to-be-detected image, determine image edge information in the to-be-detected image according to the image gradient, and generate an edge feature image corresponding to the to-be-detected image according to the image edge information.

[0087] Step 206: Determine first gray scale information corresponding to the to-be-detected image, and calculate first histogram features corresponding to the to-be-detected image according to the first gray scale information.

[0088] Step 208: Determine second gray scale information corresponding to the edge feature image, and calculate second histogram features corresponding to the feature region image according to the second gray scale information.

[0089] Step 210: Determine an image correlation degree between the to-be-detected image and the feature region image according to the first histogram features and the second histogram features.

[0090] Step 212: Determine content attribute information corresponding to the feature region image based on the image correlation degree, and generate a see-through printing detection result corresponding to the to-be-detected image according to the content attribute information.

[0091] Corresponding to the above method embodiments, the specification also provides image-based see-through printing detection device embodiments, Figure 3 A structural schematic diagram of an image-based see-through printing detection device provided in an embodiment of the specification is shown. As shown in the figure, Figure 3 The device includes: A detection module 302 configured to determine a to-be-detected image corresponding to a target document, perform edge detection on the to-be-detected image, and obtain an edge feature image corresponding to the to-be-detected image; A calculation module 304 configured to determine a feature region image based on the edge feature image, calculate first histogram features corresponding to the to-be-detected image, and calculate second histogram features corresponding to the feature region image; A determination module 306 configured to determine an image correlation degree between the to-be-detected image and the feature region image according to the first histogram features and the second histogram features, and determine content attribute information corresponding to the feature region image based on the image correlation degree. The generating module 308 is configured to generate a see-through detection result corresponding to the image to be detected according to the content attribute information.

[0092] Optionally, the detecting module 302 is further configured to determine an initial image to be detected corresponding to the target document; and perform noise reduction processing on the initial image to be detected to obtain the image to be detected corresponding to the target document.

[0093] Optionally, the detecting module 302 is further configured to perform edge detection on the image to be detected to determine an image gradient of the image to be detected; determine image edge information in the image to be detected according to the image gradient; and generate an edge feature image corresponding to the image to be detected according to the image edge information.

[0094] Optionally, the calculating module 304 is further configured to determine an image detection strategy corresponding to the image to be detected; in a case where the image detection strategy is a single-side detection strategy, take the edge feature image as a feature region image; and in a case where the image detection strategy is a double-side detection strategy, split the edge feature image to obtain a plurality of edge feature sub-images and take the plurality of edge feature sub-images as feature region images.

[0095] Optionally, the calculating module 304 is further configured to, in a case where the image detection strategy is a single-side detection strategy, determine first gray scale information corresponding to the image to be detected, calculate a first histogram feature corresponding to the image to be detected according to the first gray scale information, determine second gray scale information corresponding to the edge feature image, and calculate a second histogram feature corresponding to the feature region image according to the second gray scale information.

[0096] Optionally, the calculating module 304 is further configured to extract edge point features in the edge feature image; split the edge feature image by using the edge point features to obtain a plurality of edge feature sub-images and take the plurality of edge feature sub-images as feature region images.

[0097] Optionally, the calculating module 304 is further configured to determine a plurality of images to be detected corresponding to the plurality of edge feature sub-images in the image to be detected; determine first gray scale sub-information corresponding to the plurality of images to be detected, calculate a plurality of first histogram sub-features corresponding to the plurality of images to be detected according to the first gray scale sub-information; determine second gray scale sub-information corresponding to the plurality of edge feature sub-images, and calculate a plurality of second histogram sub-features corresponding to the plurality of edge feature sub-images according to the second gray scale sub-information.

[0098] Optionally, the determining module 306 is further configured to determine image sub-correlation degrees between the plurality of to-be-detected sub-images and the plurality of edge feature sub-images according to the first histogram sub-feature and the second histogram sub-feature; and determine the image correlation degree between the to-be-detected image and the feature region image according to the image sub-correlation degrees.

[0099] Optionally, the determining module 306 is further configured to determine a target edge feature sub-image in the plurality of edge feature sub-images and a target to-be-detected sub-image corresponding to the target edge feature sub-image according to the see-through detection result; perform image optimization on the target to-be-detected sub-image to obtain an optimized target to-be-detected sub-image; and generate a target image according to the optimized target to-be-detected sub-image.

[0100] The above is a schematic scheme of the image-based see-through detection device of the embodiment. It should be noted that the technical scheme of the image-based see-through detection device and the technical scheme of the image-based see-through detection method described above belong to the same concept, and the details of the technical scheme of the image-based see-through detection device which are not described in detail can be referred to the description of the technical scheme of the image-based see-through detection method.

[0101] Figure 4 A structural block diagram of a computing device 400 according to an embodiment of the present specification is shown. The components of the computing device 400 include but are not limited to a memory 410 and a processor 420. The processor 420 is connected with the memory 410 through a bus 430, and a database 450 is used to save data.

[0102] The computing device 400 also includes an access device 440 that enables the computing device 400 to communicate via one or more networks 460. Examples of these networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or combinations of such networks, such as the Internet. The access device 440 can include one or more of any type of network interface (for example, a network interface card (NIC)), such as an IEEE 802.11 Wireless Local Area Network (WLAN) wireless interface, a Worldwide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a Bluetooth interface, a Near Field Communication (NFC).

[0103] In one embodiment of the present specification, the above-mentioned components of the computing device 400 and other components not shown in the Figure 4 may be connected to each other, such as through a bus. It should be understood that Figure 4 The computing device structure diagram shown is only for the purpose of example, and is not a limitation on the scope of the present specification. Those skilled in the art can add or replace other components as needed.

[0104] The computing device 400 can be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (for example, a tablet computer, a personal digital assistant, a laptop computer, a notebook computer, a netbook, and the like), a mobile phone (for example, a smartphone), a wearable computing device (for example, a smart watch, smart glasses, and the like), or other types of mobile devices, or a stationary computing device such as a desktop computer or a personal computer (PC). The computing device 400 can also be a mobile or stationary server.

[0105] The processor 420 is configured to execute computer-executable instructions, which, when executed by the processor, implement the steps of the above-mentioned image-based see-through detection method.

[0106] The above is a schematic scheme of the computing device of the embodiment. It should be noted that the technical scheme of the computing device and the technical scheme of the image-based see-through printing detection method described above belong to the same concept, and details of the technical scheme of the computing device that are not described in detail can be referred to the description of the technical scheme of the image-based see-through printing detection method.

[0107] An embodiment of the present specification further provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the image-based see-through printing detection method.

[0108] The above is a schematic scheme of the computer-readable storage medium of the embodiment. It should be noted that the technical scheme of the storage medium and the technical scheme of the image-based see-through printing detection method described above belong to the same concept, and details of the technical scheme of the storage medium that are not described in detail can be referred to the description of the technical scheme of the image-based see-through printing detection method.

[0109] An embodiment of the present specification further provides a computer program product comprising a computer program or instructions, which, when executed by a processor, implement the steps of the image-based see-through printing detection method.

[0110] The above is a schematic scheme of the computer program product of the embodiment. It should be noted that the technical scheme of the computer program product and the technical scheme of the image-based see-through printing detection method described above belong to the same concept, and details of the technical scheme of the computer program product that are not described in detail can be referred to the description of the technical scheme of the image-based see-through printing detection method.

[0111] The above describes specific embodiments of the present specification. Other embodiments are within the scope of the appended claims. In some cases, the acts or steps recited in the claims can be performed in a different order than the order described in the embodiments and still achieve desirable results. In addition, the processes depicted in the figures do not necessarily require the particular order shown or sequential order in order to achieve the desired results. In some implementations, multitasking and parallel processing can be advantageous.

[0112] The computer readable medium can include any entity or apparatus capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, software distribution medium, etc. It should be noted that the computer readable medium can include appropriate additions or subtractions according to the requirements of patent practice. For example, according to the patent practice in some regions, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0113] It should be noted that, for the foregoing method embodiments, in order to facilitate description, each is described as a combination of a series of acts, but those skilled in the art should appreciate that the embodiments of the present specification are not limited by the order of the described acts, because according to the embodiments of the present specification, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should appreciate that the embodiments described in the specification are all preferred embodiments, and the acts and modules involved are not necessarily essential to the embodiments of the present specification.

[0114] In the above embodiments, the description of each embodiment has its own focus, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0115] The preferred embodiments of the present specification disclosed above are only used to help explain the present specification. The alternative embodiments do not describe all the details and do not limit the invention to the specific embodiments described. Obviously, according to the content of the embodiments of the present specification, many modifications and changes can be made. The present specification selects and describes these embodiments in order to better explain the principles and practical applications of the embodiments of the present specification, so that those skilled in the art can well understand and use the present specification.

Claims

1. An image-based print-through detection method, characterized in that: include: Determine an image to be detected corresponding to the target document, perform edge detection on the image to be detected, and obtain an edge feature image corresponding to the image to be detected; Determine a feature region image based on the edge feature image, and calculate a first histogram feature corresponding to the image to be detected and a second histogram feature corresponding to the feature region image; determining an image correlation degree between the image to be detected and the feature region image according to the first histogram feature and the second histogram feature, and determining content attribute information corresponding to the feature region image based on the image correlation degree; A print-through detection result corresponding to the image to be detected is generated according to the content attribute information.

2. The method according to claim 1, characterized in that Determine the image to be detected corresponding to the target document, including: Determine the initial image to be detected corresponding to the target document; The initial image to be detected is subjected to noise reduction processing to obtain the image to be detected corresponding to the target document.

3. The method according to claim 1, characterized in that Performing edge detection on the image to be detected to obtain an edge feature image corresponding to the image to be detected, including: Performing edge detection on the image to be detected to determine the image gradient of the image to be detected; Image edge information is determined in the image to be detected according to the image gradient, and an edge feature image corresponding to the image to be detected is generated according to the image edge information.

4. The method according to claim 1, wherein Determining a feature region image based on the edge feature image includes: Determining an image detection strategy corresponding to the image to be detected; In the case where the image detection strategy is a single-side detection strategy, the edge feature image is used as a feature area image; In the case where the image detection strategy is a double-sided detection strategy, the edge feature image is split to obtain a plurality of edge feature sub-images as feature region images.

5. The method according to claim 4, characterized in that Calculating a first histogram feature corresponding to the image to be detected and a second histogram feature corresponding to the feature area image, including: In a case where the image detection strategy is a single-side detection strategy, determining first grayscale information corresponding to the image to be detected, and calculating a first histogram feature corresponding to the image to be detected based on the first grayscale information; Determine second grayscale information corresponding to the edge feature image, and calculate second histogram features corresponding to the feature region image based on the second grayscale information.

6. The method according to claim 4, characterized in that The edge feature image is split to obtain a plurality of edge feature sub-images as feature region images, including: Extracting edge point features from the edge feature image; The edge feature image is split using the edge point features to obtain a plurality of edge feature sub-images as feature region images.

7. The method according to claim 4, characterized in that Calculating a first histogram feature corresponding to the image to be detected and a second histogram feature corresponding to the feature area image, including: Determining, in the image to be detected, a plurality of sub-images to be detected corresponding to the plurality of edge feature sub-images; Determining first grayscale sub-information corresponding to the plurality of sub-images to be detected, and calculating a plurality of first histogram sub-features corresponding to the plurality of sub-images to be detected based on the first grayscale sub-information; Second grayscale sub-information corresponding to the plurality of edge feature sub-images is determined, and a plurality of second histogram sub-features corresponding to the plurality of edge feature sub-images is calculated based on the second grayscale sub-information.

8. The method according to claim 7, characterized in that Determining an image correlation degree between the image to be detected and the feature area image according to the first histogram feature and the second histogram feature includes: determining, based on the first histogram sub-feature and the second histogram sub-feature, image sub-correlations between the plurality of sub-images to be detected and the plurality of edge feature sub-images; An image correlation degree between the image to be detected and the feature region image is determined according to the image sub-correlation degree.

9. The method according to claim 8, characterized in that After generating a print-through detection result corresponding to the image to be detected according to the content attribute information, the method further includes: determining a target edge feature sub-image and a target sub-image to be detected corresponding to the target edge feature sub-image from the plurality of edge feature sub-images according to the print-through detection result; performing image optimization on the target to-be-detected sub-image to obtain an optimized target to-be-detected sub-image; A target image is generated according to the optimized target sub-image to be detected.

10. An image-based print-through detection device, characterized in that: include: A detection module is configured to determine an image to be detected corresponding to a target document, perform edge detection on the image to be detected, and obtain an edge feature image corresponding to the image to be detected; a calculation module configured to determine a feature region image based on the edge feature image, and calculate a first histogram feature corresponding to the image to be detected, and a second histogram feature corresponding to the feature region image; a determination module configured to determine an image correlation degree between the image to be detected and the feature region image according to the first histogram feature and the second histogram feature, and determine content attribute information corresponding to the feature region image based on the image correlation degree; The generating module is configured to generate a print-through detection result corresponding to the image to be detected according to the content attribute information.

11. A computing device, characterized in that include: memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the image-based print-through detection method according to any one of claims 1 to 9 are implemented.

12. A computer-readable storage medium, characterized in that It stores computer-executable instructions, which, when executed by a processor, implement the steps of the image-based print-through detection method according to any one of claims 1 to 9.

13. A computer program product, characterized in that The method comprises a computer program or instructions, which, when executed by a processor, implements the steps of the image-based print-through detection method according to any one of claims 1 to 9.

Images