A screen dot matrix watermark identification method and system

Abstract

The present application relates to the technical field of screen watermark, and provides a screen dot matrix watermark identification method and system, which comprises the following steps: based on a pre-processed screen watermark image, a dot matrix binary image and a dot matrix center contour are obtained through dot matrix extraction; in the dot matrix binary image, a rectangular region is cut out based on the contour of each dot, and it is judged whether the feature of each rectangular region meets the starting dot matrix feature; if yes, the coordinates, dot width and dot spacing of the starting dot matrix are recorded, and searching is started from the coordinates of the starting dot matrix until a character dot matrix is found; based on the dot width and dot spacing, the character dot matrix is segmented to obtain a plurality of predicted dots, and the area coincidence degree of each predicted dot and an actual dot is calculated; based on the area coincidence degree, the binary information of the character dot matrix is calculated, and the character is mapped through a code table. To some extent, the local watermark is prevented from being damaged and blocked, and the anti-interference performance and extraction efficiency are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of screen watermark, and particularly relates to a screen dot matrix watermark identification method and system. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute the prior art.

[0003] With the development of science and technology, the means of leakage is also becoming more and more rich. In particular, the popularization of intelligent devices, such as smart phones, watches and other electronic devices, the above-mentioned devices with shooting function can make users more convenient to shoot the desired material. Through the way of secretly taking pictures, screenshots and other ways of computer screen, the confidential information is spread, which brings huge influence and economic loss to enterprises or individuals.

[0004] Screen watermark is to add watermark information on the computer screen, which can protect the web page, chat record, engineering drawing and other contents displayed on the screen, and can achieve the purpose of protecting copyright and privacy. When the screen content is leaked through screenshot or photograph, the watermark information in the photo or screenshot can be used to trace the leakage machine, the leakage personnel and the leakage time.

[0005] Screen watermark usually contains dot matrix watermark and color block watermark. In the existing screen watermark identification scheme, the method based on deep learning is usually used for identification, such as watermark identification based on conditional generative adversarial neural network (cGAN) and watermark identification based on semantic segmentation (U-Net). Figure 1

[0006] The U-Net network uses large-scale data samples to train the model, and through layer-by-layer sampling, it captures features at different levels and integrates them through feature superposition. This method can eliminate the content other than the encrypted information through the downsampling process, and can extract the screen watermark information at different resolutions through the multi-layer network structure, and then fuse them through the up-sampling convolution to obtain the black and white alternating watermark image. For screen watermark screenshot, this method has strong anti-interference ability and accurate identification result.

[0007] Similar to the deep learning method based on U-Net network, the recognition effect is based on large-scale data samples. The recognition effect of screen watermark screenshot is good, but the recognition effect of photograph is not good, and there are the following defects:

[0008] ​a) More practical application scenarios of screen watermark identification are to identify the photographed images of screen watermarks. Due to the influence of light, shooting angle, moire, screen resolution and other factors, it is difficult to cover all shooting scenarios during model training. A large amount of samples of different scenarios need to be collected for training with the help of high-performance servers, which is a large amount of work and often cannot meet the requirements of practical applications.

[0009] b) Moire has a large interference on model training. The actual trained model is greatly affected by the severity of moire. The images photographed by different mobile phones have large differences in moire. The influence of moire needs to be removed during model training. The severity of moire is different in different areas of the same image. For screen dot watermarks, the dot watermarks in the photographed image are easily fused with moire. After removing moire, the watermark information may also be removed, and the recognition effect is not good in practical application. SUMMARY

[0010] In order to solve the technical problems in the background art, the present application provides a screen dot watermark identification method and system. In view of the characteristics that the size of the dot in the photographed dot array, the dot distance and the distance between the dot arrays are not fixed and are not proportional, the coincidence degree of the area of the predicted dot and the actual dot is calculated through the initial dot array identification, the character dot array segmentation, and the area of the predicted dot and the actual dot, which to some extent avoids the local watermark from being damaged and blocked, and improves the anti-interference and extraction efficiency.

[0011] In order to achieve the above purpose, the present application adopts the following technical solutions:

[0012] The first aspect of the present application provides a screen dot watermark identification method, which comprises:

[0013] obtaining a screen watermark image;

[0014] based on the different acquisition scenarios and the different moire intensities of the screen watermark image, different pre-processing is performed on the screen watermark image to obtain a pre-processed screen watermark image;

[0015] based on the pre-processed screen watermark image, dot array extraction is performed to obtain a dot array binary image and the contour of the dot in the dot array;

[0016] in the dot array binary image, a rectangular area is cropped based on the contour of each dot, and it is judged whether the characteristics of each rectangular area meet the characteristics of the initial dot array. If yes, the coordinates, dot width and dot spacing of the initial dot array are recorded, and searching is started from the coordinates of the initial dot array until a character dot array is found. Based on the dot width and dot spacing, the character dot array is segmented to obtain a plurality of predicted dots, and the coincidence degree of the area of each predicted dot and the actual dot is calculated. Based on the coincidence degree, the binary information of the character dot array is calculated, and the character is mapped through a code table.

[0017] Further, the features of the rectangular region include: dot array number, row number, column number, dot width, dot height and dot spacing.

[0018] Further, the specific steps of the dot array extraction include: after the pre-processed screen watermark image is sequentially subjected to gray image extraction, noise reduction and histogram equalization, a dot array binary image is obtained through an edge detection algorithm; and in the dot array binary image, the contour of the dot in the dot array is recognized and labeled.

[0019] Further, if the acquisition scenario of the screen watermark image is screen capture, the pre-processing includes sharpening, erosion, binarization and histogram equalization.

[0020] Further, if the acquisition scenario of the screen watermark image is screen capture, and the moire is determined to be mild moire, the pre-processing includes mean filtering.

[0021] Further, if the acquisition scenario of the screen watermark image is screen capture, and the moire is determined to be mild moire, the pre-processing includes mean filtering.

[0022] Further, if the acquisition scenario of the screen watermark image is screen capture, the pre-processing includes binarization, sharpening and twice erosion.

[0023] The second aspect of the application provides a screen dot array watermark identification system, which comprises:

[0024] a data acquisition module configured to acquire a screen watermark image;

[0025] a pre-processing module configured to, based on different acquisition scenarios of the screen watermark image and different moire strengths, perform different pre-processing on the screen watermark image to obtain a pre-processed screen watermark image;

[0026] a dot array extraction module configured to, based on the pre-processed screen watermark image, perform dot array extraction to obtain a dot array binary image and a contour of a dot in the dot array;

[0027] a watermark information extraction module configured to, in the dot array binary image, cut out a rectangular region based on the contour of each dot, and determine whether the features of each rectangular region meet the features of a starting dot array; if yes, record the coordinates, dot width and dot spacing of the starting dot array, and start searching from the coordinates of the starting dot array until a character dot array is found, segment the character dot array based on the dot width and dot spacing to obtain a plurality of predicted dots, calculate the area overlap degree of each predicted dot and an actual dot, calculate the binary information of the character dot array based on the overlap degree, and map the binary information to characters through a code table.

[0028] The third aspect of the present application provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the steps of the screen dot matrix watermark identification method.

[0029] The fourth aspect of the present application provides a computer device, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the screen dot matrix watermark identification method when executing the program.

[0030] Compared with the prior art, the present application has the following beneficial effects:

[0031] The present application is aimed at the characteristics that the size of the dot in the screen shot dot matrix, the dot distance and the distance between dot matrices are not fixed and not proportional. Through the initial dot matrix identification, the character dot matrix segmentation, the area coincidence calculation of the predicted dot and the actual dot, the local watermark is avoided to be damaged and blocked to a certain extent, and the anti-interference and the extraction efficiency are improved.

[0032] The present application creatively combines the dot matrix identification self-adaptation and the method of identifying multiple dot matrix blocks and correcting the results to further improve the deficiencies of the prior art, improve the watermark extraction efficiency and meet the demand of simultaneously identifying the screen shot watermark and the screen shot watermark in the actual application scenario. BRIEF DESCRIPTION OF DRAWINGS

[0033] The drawings accompanying the specification of the present application serve to provide a further understanding of the present application, and the illustrative embodiments of the present application and the description thereof serve to explain the present application, and do not constitute an improper limitation of the present application.

[0034] Figure 1 is a flowchart of a screen dot matrix watermark identification method of the first embodiment of the present application;

[0035] Figure 2 is a dot matrix extraction flowchart of the first embodiment of the present application;

[0036] FIG. 3(a) is a starting dot matrix detection flowchart of the first embodiment of the present application;

[0037] FIG. 3(b) is a character recognition flowchart of the first embodiment of the present application;

[0038] Figure 4 is a screen shot with dot matrix watermark of the first embodiment of the present application;

[0039] Figure 5 is a dot matrix watermark binary image of the first embodiment of the present application;

[0040] Figure 6 is a starting dot matrix schematic diagram of the first embodiment of the present application;

[0041] Fig. 7(a) is a first time screen shot binary image of the dot matrix of the embodiment one of the present application;

[0042] Fig. 7(b) is a second time screen shot binary image of the dot matrix of the embodiment one of the present application;

[0043] Figure 8 Fig. 8 is a schematic diagram of the dot matrix block of the embodiment one of the present application;

[0044] Figure 9 Fig. 9 is a schematic diagram of the inter-dot matrix distance of the embodiment one of the present application;

[0045] Figure 10 Fig. 10 is a schematic diagram of the inter-dot matrix distance of the embodiment one of the present application;

[0046] Figure 11 Fig. 11 is a schematic diagram of the dot matrix code representation of the embodiment one of the present application;

[0047] Figure 12 Fig. 12 is a schematic diagram of the dot matrix information fusion of the embodiment one of the present application. DETAILED DESCRIPTION

[0048] The present application will be further described below in conjunction with the accompanying drawings and embodiments.

[0049] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0050] Embodiment one

[0051] The embodiment provides a screen dot matrix watermark identification method.

[0052] The screen dot matrix watermark identification method provided by the embodiment is based on OpenCV.

[0053] The screen dot matrix watermark identification method provided by the embodiment can quickly and accurately identify the two kinds of screen dot matrix watermarks of screen shot and screen shot, and has practical application value.

[0054] In the present application, the image processing method based on OpenCV is used to preprocess the screen watermark image. After preprocessing, the screen watermark image is converted into a black and white binary image with dot matrix watermark. According to the difference between the screen capture and the screen shot, different preprocessing methods are summarized. For screen capture, the methods of sharpening, erosion, binarization and histogram equalization are used. For screen shot, different filtering algorithms are selected according to the different intensity of moire. For light moire, mean filtering is used, and for heavy moire, the method of bilateral filtering + mean filtering is used. In order to process the screen shot which may not be clear and the dot matrix is small, the method of binarization, sharpening and twice erosion is used to enhance the dot matrix watermark in the preprocessing method, which is convenient for subsequent extraction.

[0055] Among them, light and heavy moire are processed by two ways: ① there are light and heavy moire processing in the user interface preprocessing operation, and the user selects when manually extracting; ② there is an automatic extraction function in the program, which preprocesses the same screen watermark image with light and heavy moire respectively, then extracts the watermark information and fuses it.

[0056] The preprocessing method based on OpenCV classifies the screen watermark image according to the scene, so that different preprocessing methods are used, without the need to collect a large number of samples to train the model. For the problem of moire in screen shot, different filtering algorithms are used to denoise the watermark image according to the different intensity of moire, then the denoised images are extracted respectively, and finally the extraction results are fused together and corrected to obtain the watermark information.

[0057] Specifically, the screen dot matrix watermark recognition is to extract the watermark information from the image with dot matrix watermark, which mainly includes two steps:

[0058] ① Dot matrix extraction, that is, to strip the binary image of dot matrix from the watermark image containing dot matrix, that is, only the black and white binary image with dot matrix;

[0059] ② Watermark information extraction, to extract the watermark information from the binary image of dot matrix.

[0060] Among them, the dot matrix extraction process is as shown in Figure 2 , including: image preprocessing; Canny edge detection; OpenCV contour recognition and contour drawing.

[0061] The general image preprocessing includes extracting a gray image, noise reduction, and histogram equalization to enhance contrast; a Canny edge detection algorithm detects rectangular dot arrays in the image and labels (here, referring to labeling the edge profile of each point in the dot array); OpenCV (open source computer vision library) contour extraction findContours identifies the profile coordinates of the labeled dot array block (here, referring to labeling the profile coordinates of each point in the dot array); the profile coordinates are used to draw a frame to visualize the dot array, and finally a dot array binary image and corresponding profile coordinates are obtained.

[0062] The watermark information extraction process is shown in FIGS. 3(a) and 3(b) and includes start dot array detection, character recognition, and result fusion.

[0063] (1) As shown in FIG. 3(a), the start dot array detection includes scanning the profile of each point in all dot arrays to calculate the upper left corner and the lower right corner of each point in the dot array; for each point, the upper left corner is taken as the origin to extend a certain distance to the lower right corner for detection, and an ROI rectangular region is cropped out to determine whether the number of points, rows and columns, dot array width (point width), dot height, and dot spacing (row and column spacing between points) in the rectangular region meet the characteristics of the start dot array, and if so, the left upper corner coordinates of the start dot array, the dot width (average value of all dot widths in the start dot array), and the dot spacing (average value of all dot spacings in the start dot array) are output.

[0064] Each point in the dot array is a small rectangle, and the OpenCV contour recognition algorithm will give a set of coordinates, which are on the four sides of the rectangle, but not the vertices of the sides. Therefore, the upper left corner and the lower right corner of each matrix need to be calculated (x-axis extends to the right, y-axis extends downward, and the minimum values of x and y in the set of points are the upper left corner, and the maximum values of x and y are the lower right corner).

[0065] In the ROI rectangular region cropping process, first, the left upper corner of the point contour is taken as the origin, and the point width is taken as the edge length to crop a rectangular region, and it is judged whether the rectangular region satisfies the starting point array feature; if not, the edge length is changed to 2*point width, a rectangular region is cropped, and it is judged whether the rectangular region satisfies the starting point array feature; if not, the edge length is increased by one point width again, the rectangular region is continuously cropped and judged, and the like is sequentially repeated until the starting point array is found. The features of the starting point array are that the number of points is 9 points, the number of rows and the number of columns are both 3, the point width is similar (that is, the difference between all point widths is less than a threshold value), the point height is similar (that is, the difference between all point heights is less than a threshold value), the row spacing between points is similar (that is, the difference between all row spacings is less than a threshold value), and the column spacing between points is similar (that is, the difference between all column spacings is less than a threshold value). Because the screen shot may introduce noise points, the noise points may be large or small, and the size judgment can filter most of the noise points. If only the size and number of points are judged, when the size of the noise point is similar to the size of the point in the point array, it may be misidentified as a starting point array. For example, the character point array, after adding noise points, also satisfies the feature of 9 points in terms of quantity, but the position of the noise point is random, and the row and column position information often does not satisfy it, so the above features can avoid misidentification.

[0066] Specifically, Figure 4 is a screenshot with a point array watermark. The point array watermark binary image after image preprocessing and enhancement is as shown in Figure 5

[0067] Among them, the starting point array is a complete 3*3 point array block, which does not represent a specific character, and identifies the beginning of the point array, as shown in Figure 6 The number of point arrays is 9, the number of rows and columns is 3, and the point array width is 3*point width+2*point distance.

[0068] The detection of the starting point array is to find the coordinates of all starting point arrays in the point array watermark binary image. The screen shot may be affected by the distance of screen shooting, the screen shooting angle, and the like, and the size of the point and the distance are not fixed, as shown in FIG. 7(a) and FIG. 7(b).

[0069] FIG. 7(a) and FIG. 7(b) are point array binary images obtained by preprocessing the same point array after two screen shots, and it can be seen that the size of the point in the point array and the distance between the points are quite different. The detection of the starting point array is to obtain the coordinates of the starting point array in each image, the actual point size and distance information.

[0070] ​(2) As shown in Figure 3(a), character recognition includes: searching to the right based on the coordinates of the starting dot matrix, finding a complete dot matrix (5 points) to obtain the distance between the dots, dividing the dot matrix into blocks according to the dot matrix size and distance, and calculating the overlap between the predicted point and the actual point area in a single dot matrix block (a certain point) in turn. If the overlap is greater than the threshold, it is recorded as "1" and if it is less than the threshold, it is recorded as "0". In this way, the binary corresponding to a single dot matrix is ​​obtained. The binary result is mapped to the corresponding character result to obtain the character information of a single dot matrix.

[0071] The search to the right involves probing to the right at a unit distance, which is 1 / 4 of the dot matrix width. Based on the coordinates of the starting dot matrix (located at the top left corner), a small rectangular area is divided to the right at the top right corner coordinates of the starting dot matrix, with each unit distance representing a unit distance. It is then checked whether this area contains a complete character dot matrix (characterized by 5 dots, similar width and height, similar row and column distances, and similar spacing to the center point of the starting dot matrix). If not, the distance is incremented by one unit, and the rectangular area is divided to the right again, with the process repeated until a complete character dot matrix is ​​found. Finally, the distance between the dots is calculated.

[0072] Specifically, the information-carrying dot matrix block is a 3x3 dot matrix block: the starting dot matrix is ​​in the upper left corner, and the other dot matrices represent character information. For example... Figure 8 As shown.

[0073] After obtaining the coordinates of the starting dot matrix, search to the right for the character dot matrix (the character dot matrix has 5 dots, and the starting dot matrix has 9 dots). The distance between the starting dot matrix and the character dot matrix is ​​the distance between the dot matrices. Figure 9 As shown.

[0074] After obtaining the distances between the dot matrix blocks, the dot matrix blocks can be segmented based on the previously obtained information such as the starting dot matrix coordinates, dot size, and distances. Each dot matrix block is divided into 9 smaller blocks: the first is the starting dot matrix, and the other 8 are character dot matrices. The characters represented by these 8 character dot matrices are then identified sequentially. The process of identifying the character dot matrix is ​​as follows: Figure 10 As shown.

[0075] Based on the obtained information such as distance and size, the coordinates of a point in the dot matrix can be calculated, which is the predicted position of the point. Figure 10 The dark outline represents the actual position of the points in the dot matrix, while the white outline represents the actual position of the points in the matrix. By calculating the overlap between the predicted points and the actual points, such as setting the overlap threshold to 70% (meaning the overlap of the areas of two rectangles is not less than 70% is recorded as 1, otherwise as 0), the binary information represented by the dot matrix can be identified. Figure 10 The binary information represented by the dot matrix of Chinese characters is 001010111.

[0076] The binary information is mapped to characters through a code table, and the code table has 36 characters, respectively representing numbers 0-9, letters a-z, and specifically as shown in Figure 11 (only numbers and the first five characters are listed in the figure).

[0077] (3) Result fusion: finally, the character information of multiple dot matrixes is fused, that is, the character with the largest number of occurrences is selected for each bit to obtain the final recognition result.

[0078] Specifically, because there are multiple dot matrix blocks in a dot matrix watermark image, due to occlusion, image preprocessing, etc., the information in the dot matrix block cannot be completely recognized, and only part of the character can be recognized, as shown in Figure 12

[0079] The information of the two dot matrix blocks is 00000XX1 and 000X0t3X, respectively, and the uppercase X represents that the position is not recognized. The two recognition results can be fused, and the recognized dot matrix information is taken out, and finally the recognition result of the dot matrix is 00000t31.

[0080] The present application provides a screen dot matrix watermark recognition method based on OpenCV, which can recognize the watermark image of screen capture and screen shooting at the same time, has strong anti-interference ability, and has practical application value.

[0081] In view of the characteristics that the size of the dot in the dot matrix, the distance between the dots, and the distance between the dot matrices in the screen shooting scene are not fixed and are not proportional, a dot matrix adaptive recognition algorithm is designed, which can avoid local watermark damage and occlusion to a certain extent, and improve the anti-interference ability and extraction efficiency. The recognition results of multiple dot matrix blocks are fused and corrected, which can improve the accuracy of the recognition results.

[0082] The present application provides a screen dot matrix watermark recognition method based on OpenCV, which creatively combines the dot matrix recognition adaptive method and the method of recognizing multiple dot matrix blocks and fusing and correcting the results, to further improve the shortcomings of the prior art, improve the watermark extraction efficiency, and meet the demand of recognizing screen capture watermark and screen shooting watermark at the same time in actual application scenarios.

[0083] Example two

[0084] The present application provides a screen dot matrix watermark recognition system, comprising:

[0085] A data acquisition module configured to acquire a screen watermark image;

[0086] A preprocessing module configured to perform different preprocessing on the screen watermark image based on different acquisition scenarios and different moire intensities of the screen watermark image to obtain a preprocessed screen watermark image; ​

[0087] a dot matrix extraction module configured to obtain a dot matrix binary image and a contour of a dot matrix center based on the preprocessed screen watermark image through dot matrix extraction;

[0088] a watermark information extraction module configured to crop a rectangular region based on the contour of each dot in the dot matrix binary image, and determine whether a feature of each rectangular region satisfies a starting dot matrix feature, if yes, record a coordinate of the starting dot matrix, a dot width and a dot pitch, and start searching from the coordinate of the starting dot matrix until a character dot matrix is found, segment the character dot matrix based on the dot width and the dot pitch to obtain a plurality of predicted dots, calculate an area overlap degree of each predicted dot and an actual dot, calculate binary information of the character dot matrix based on the overlap degree, and map the binary information into a character through a code table.

[0089] It should be noted that each module in the embodiment corresponds to each step in Embodiment One, and the specific implementation process is the same, which will not be repeated here.

[0090] Embodiment Three

[0091] The embodiment provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to realize the steps in the screen dot matrix watermark identification method according to Embodiment One.

[0092] Embodiment Four

[0093] The embodiment provides a computer device, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor realizes the steps in the screen dot matrix watermark identification method according to Embodiment One when executing the program.

[0094] The present application is described with reference to flowcharts and / or block diagrams of the method, device (system) and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be realized by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a device for realizing the functions specified in the flowcharts and / or block diagrams. Figure 1 The function of one flow or multiple flows and / or blocks Figure 1 The device for realizing the function of one block or multiple blocks.

[0095] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are generated to realize the computer-implemented processes in the computer or other programmable devices, and the instructions executed in the computer or other programmable devices provide a process for implementing the functions specified in the flowchart Figure 1 or multiple flows and / or blocks Figure 1 or multiple blocks in the flowchart or multiple blocks in the flowchart.

Claims

1. A method for screen dot pattern watermark identification, characterized in that, The method comprises the following steps: acquiring a screen watermark image; based on the acquisition scene and the Moire intensity of the screen watermark image, different pre-processing is performed on the screen watermark image to obtain a pre-processed screen watermark image, specifically: if the acquisition scene of the screen watermark image is screen capture, the pre-processing includes sharpening, erosion, binarization and histogram equalization; if the acquisition scene of the screen watermark image is screen shooting and the Moire is determined to be mild Moire, the pre-processing includes mean filtering; if the acquisition scene of the screen watermark image is screen shooting and the Moire is determined to be severe Moire, the pre-processing includes bilateral filtering and mean filtering; if the acquisition scene of the screen watermark image is screen shooting, the pre-processing includes binarization, sharpening and twice erosion; based on the pre-processed screen watermark image, dot matrix extraction is performed to obtain a dot matrix binary image and the contour of a dot matrix center point; in the dot matrix binary image, a rectangular region is cropped based on the contour of each dot, and it is determined whether the feature of each rectangular region meets the starting dot matrix feature; if it meets, the coordinates, dot width and dot spacing of the starting dot matrix are recorded, and searching is started from the coordinates of the starting dot matrix until a character dot matrix is found; based on the dot width and dot spacing, the character dot matrix is segmented to obtain a plurality of predicted points, and the area coincidence degree of each predicted point with an actual point is calculated; based on the area coincidence degree, the binary information of the character dot matrix is calculated, and the character is mapped through a code table.

2. The method of claim 1, wherein the step of identifying the screen dot pattern watermark is characterized by, The features of the rectangular region include: dot matrix number, row number, column number, dot width, dot height and dot spacing.

3. The method of claim 1, wherein the step of embedding the watermark comprises the steps of: generating a watermark pattern; and embedding the watermark pattern in the screen dot pattern. The specific steps of the dot matrix extraction include: after the pre-processed screen watermark image is sequentially subjected to gray image extraction, noise reduction and histogram equalization, a dot matrix binary image is obtained through an edge detection algorithm; in the dot matrix binary image, the contour of a dot matrix center point is identified and labeled.

4. A screen dot pattern watermark identification system characterized by comprising: The method comprises the following steps: a data acquisition module configured to acquire a screen watermark image; a preprocessing module configured to perform different pre-processing on the screen watermark image based on the acquisition scene and the Moire intensity of the screen watermark image to obtain a pre-processed screen watermark image, specifically: if the acquisition scene of the screen watermark image is screen capture, the pre-processing includes sharpening, erosion, binarization and histogram equalization; if the acquisition scene of the screen watermark image is screen shooting and the Moire is determined to be mild Moire, the pre-processing includes mean filtering; if the acquisition scene of the screen watermark image is screen shooting and the Moire is determined to be severe Moire, the pre-processing includes bilateral filtering and mean filtering; if the acquisition scene of the screen watermark image is screen shooting, the pre-processing includes binarization, sharpening and twice erosion; a dot matrix extraction module configured to perform dot matrix extraction based on the pre-processed screen watermark image to obtain a dot matrix binary image and the contour of a dot matrix center point; The watermark information extraction module is configured to: in the dot matrix binary image, a rectangular region is cropped based on the contour of each dot, and it is judged whether the feature of each rectangular region satisfies the starting dot matrix feature; if yes, the coordinates, dot width and dot pitch of the starting dot matrix are recorded, and searching is started from the coordinates of the starting dot matrix until a character dot matrix is found; the character dot matrix is segmented based on the dot width and dot pitch to obtain a plurality of predicted dots, and the area coincidence degree of each predicted dot with an actual dot is calculated; the binary information of the character dot matrix is calculated based on the area coincidence degree, and is mapped into a character through a code table.

5. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the steps in the screen dot matrix watermark identification method of any one of claims 1-3.

6. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the steps in the screen dot matrix watermark identification method of any one of claims 1-3.

Images