Image security protection method and system based on multi-domain watermark dual verification

Abstract

The invention provides an image security protection method and system based on multi-domain watermark dual verification, and the method comprises the steps: firstly obtaining an original image and preset main and auxiliary watermarks, enabling the main watermark to be associated with an image source, enabling the auxiliary watermark to correspond to the image content features, carrying out the domain feature analysis of the original image, and obtaining a color feature domain and a texture feature domain, embedding the main watermark into a color feature domain, embedding the auxiliary watermark into a texture feature domain, performing watermark extraction and comparison on the feature domain containing the watermark, generating a dual verification result indicating the source legality and the content integrity of the image, and finally triggering an image security protection mechanism based on the dual verification result. Normal use of the image is allowed or an abnormal interception process is started, dynamic and effective protection of image security is realized, and reliability and practicability of image security protection are greatly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image security protection, in particular to an image security protection method and system based on multi-domain watermark double verification. BACKGROUND

[0002] In the field of image security protection, it is crucial to ensure the legality of the image source and the integrity of the image content. Existing image security protection methods have many shortcomings. On the one hand, some methods only use a single watermarking technology, such as using only image content-based digital watermarking. Although the above watermarking can reflect the characteristics of the image content to some extent, it has weak identification ability for the image source. Once the image is illegally copied or disseminated, it is difficult to accurately trace its original source, and it is difficult to effectively combat infringement. On the other hand, some methods use double watermarking, but simply embed two watermarks into the same feature domain of the image, without fully considering the characteristics and advantages of different feature domains of the image. Since the color features and texture features of the image have different properties, embedding watermarks into the same feature domain may cause difficulty in balancing the robustness and invisibility of the watermark, and when the image is attacked, the watermark is easily damaged or difficult to accurately extract, thereby affecting the verification effect of the legality of the image source and the integrity of the image content. In addition, after verifying the watermark, the existing methods lack effective security protection mechanisms and cannot take appropriate measures in a timely manner based on the verification results to protect the safe use of the image, making it difficult to meet the increasingly complex image security protection needs. SUMMARY

[0003] In view of the above-mentioned problems, in combination with the first aspect of the present application, the embodiments of the present application provide an image security protection method based on multi-domain watermark double verification, which comprises: Obtaining an original image and preset primary watermark and secondary watermark, the original image containing a pixel color array and a spatial texture layout, the primary watermark being an identification sequence associated with the image source, and the secondary watermark being a feature code corresponding to the image content; Performing domain feature analysis on the original image to obtain a color feature domain and a texture feature domain of the original image, the color feature domain being composed of pixel value distribution of each color channel, and the texture feature domain being composed of texture structure patterns in different regions of the image; Embedding the primary watermark into the color feature domain and embedding the secondary watermark into the texture feature domain to obtain a color feature domain containing the primary watermark and a texture feature domain containing the secondary watermark; Performing watermark extraction and comparison on the color feature domain containing the primary watermark and the texture feature domain containing the secondary watermark to generate a double verification result; Triggering an image security protection mechanism based on the double verification result, the image security protection mechanism including allowing normal use of the image or starting an abnormal interception process.

[0004] In still another aspect, the embodiments of the present application also provide an image security protection system based on multi-domain watermark double verification, comprising a processor, a machine readable storage medium, the machine readable storage medium and the processor being connected, the machine readable storage medium being used for storing programs, instructions or codes, and the processor being used for executing the programs, instructions or codes in the machine readable storage medium to realize the above method.

[0005] Based on the above aspects, the embodiments of the present application obtain an original image and preset primary watermark and secondary watermark, wherein the primary watermark is associated with image source, and the secondary watermark corresponds to image content features, for domain feature analysis on the original image to obtain a color feature domain and a texture feature domain, fully utilizing the characteristics and advantages of different feature domains of the image, pixel value distribution of the color feature domain and texture structure mode of the texture feature domain providing a more suitable environment for watermark embedding, embedding the primary watermark into the color feature domain and embedding the secondary watermark into the texture feature domain, realizing independent embedding of the primary and secondary watermarks in different feature domains, improving the robustness and invisibility of the watermark, making the watermark more difficult to be destroyed and perceived when facing various attacks, and performing watermark extraction and comparison on the color feature domain containing the primary watermark and the texture feature domain containing the secondary watermark to generate double verification results, which can more accurately and comprehensively indicate the legality of the image source and the integrity of the image content, trigger an image security protection mechanism based on the double verification results, allow normal use of the image or start an abnormal interception process according to the verification result, realize dynamic and effective protection of image security, and greatly improve the reliability and practicality of image security protection. BRIEF DESCRIPTION OF DRAWINGS

[0006] Figure 1 is an execution flow diagram of the image security protection method based on multi-domain watermark double verification provided by the embodiments of the present application.

[0007] Figure 2 is a schematic diagram of exemplary hardware and software components of the image security protection system based on multi-domain watermark double verification provided by the embodiments of the present application. DETAILED DESCRIPTION

[0008] The present application will be described in detail below with reference to the accompanying drawings, Figure 1 is a flow diagram of the image security protection method based on multi-domain watermark double verification provided by an embodiment of the present application, and the image security protection method based on multi-domain watermark double verification will be described in detail below.

[0009] Step S110: obtaining an original image and preset primary watermark and secondary watermark, the original image containing a pixel color array and a spatial texture layout, the primary watermark being an identification sequence associated with image source, and the secondary watermark being a feature code corresponding to image content.

[0010] In this embodiment, the construction process record image of a reinforced concrete frame structure in the construction field can be taken as an example, which is used to record the actual situation of the construction links such as steel bar binding and formwork installation of beams, columns, slabs and other components in the frame structure.

[0011] In order to ensure the security of the construction process record image in the process of transmission, storage and use, the original image and the preset primary watermark and auxiliary watermark matched with the original image need to be obtained. After the original image is shot by the high-definition camera at the construction site, it is sent to the image receiving module of the engineering management system through the special data transmission line. The primary watermark and the auxiliary watermark are pre-stored in the watermark database of the engineering management system, and each original image has a unique identifier, by which the corresponding primary watermark and auxiliary watermark can be accurately retrieved from the watermark database.

[0012] Step S111: receiving the original image transmitted externally, the original image existing in the form of a two-dimensional pixel matrix, each element in the two-dimensional pixel matrix representing the color information of a pixel point.

[0013] The original image transmitted externally is processed by encoding and transmitted in the form of a data packet in the transmission line. After the image receiving module of the engineering management system receives the data packet, it first checks the data packet for completeness and accuracy. If there is a data packet loss or damage, a retransmission request can be sent to the sending end.

[0014] After the verification, the image receiving module decodes the data packet to restore the encoded image data to the original digital image information, which is presented in the form of a two-dimensional pixel matrix at this time. The number of rows and columns of the two-dimensional pixel matrix is determined by the resolution of the camera. The size of the two-dimensional pixel matrix of the image shot by a camera with different resolution is different.

[0015] Each element in the two-dimensional pixel matrix corresponds to a pixel point in the image, and the color information contained in each element is related to the color model used by the image. If the RGB color model is used, each element is composed of three color components of red, green and blue, and the three components together determine the color presented by the pixel point.

[0016] Step S112: extracting a pixel color array from the two-dimensional pixel matrix of the original image, the pixel color array being composed of color values of pixel points in matrix arrangement order.

[0017] When extracting the pixel color array from the two-dimensional pixel matrix, the extraction is carried out according to the inherent arrangement order of the matrix. Specifically, the color values of each pixel point from left to right in the first row of the two-dimensional pixel matrix are extracted in sequence, and the color values are arranged in sequence.

[0018] After the color values of the pixels in the first row are extracted, the color values of the pixels in the second row are extracted in sequence from left to right, and the extracted color values are arranged in sequence behind the color values of the pixels in the first row.

[0019] In this way, until the color values of all the pixels in the last row of the two-dimensional pixel matrix are extracted, the pixel color array extracted is a one-dimensional sequence, which completely retains the original distribution order of the color information in the two-dimensional pixel matrix.

[0020] Step S113: Analyze the distribution of the pixels in the original image, determine the texture direction and density of different regions, and generate a spatial texture layout, wherein the texture characteristics of each region in the spatial texture layout are determined by the arrangement of the continuous pixels.

[0021] When analyzing the distribution of the pixels in the original image, the image is divided into a plurality of non-overlapping sub-regions in a region division manner. The size of the divided sub-regions can be adjusted according to the complexity of the texture structure in the image. For a region with complex texture structure, the sub-region can be divided smaller; for a region with simple texture structure, the sub-region can be divided larger.

[0022] For each sub-region, the rate of change of the color values of the pixels in the region is calculated, and the texture direction is determined by the size and direction of the rate of change. At the same time, the number of times of significant change of the color values of the pixels in a unit area is counted to determine the texture density of the region.

[0023] In the image recorded in the construction process of the reinforced concrete frame structure, the color values of the pixels in the edge region of the beam and column change obviously, and the texture direction is relatively regular; while the color values of the pixels in the concrete surface region change relatively gently, and the texture direction is relatively random. According to these analysis results, the continuous pixels with similar texture direction and density in the image are classified into a region, thereby generating a spatial texture layout.

[0024] Step S114: Retrieve the main watermark from the preset watermark library, wherein the identification sequence of the main watermark is arranged by specific characters according to a preset rule, and the identification sequence is associated with the source information of the original image.

[0025] The preset watermark library is a database in the engineering management system specially used for storing various types of watermark information, and each main watermark in the watermark library is associated with specific image source information. When the main watermark is retrieved, the engineering management system searches in the watermark library according to the unique identifier of the original image.

[0026] After the corresponding main watermark is retrieved, it is called from the watermark library. The identification sequence of the main watermark is composed of a plurality of specific characters, which can be letters, numbers or specific symbols. The arrangement order of the characters follows a preset rule, which is formulated by the engineering management department according to actual management needs.

[0027] For example, the identification sequence can contain characters related to information such as the construction unit code, the shooting device number, the shooting time, etc. Through the specific arrangement of the above characters, the source information of the original image can be accurately reflected.

[0028] Step S115: The auxiliary watermark is called from the preset watermark library, and the feature code of the auxiliary watermark is generated based on the content features of the original image. Each part of the feature code corresponds to a specific content feature in the original image.

[0029] Similar to calling the main watermark, the auxiliary watermark is also retrieved in the preset watermark library through the unique identifier of the original image. The feature code of the auxiliary watermark is generated based on the content features of the original image. When generating the feature code, each content feature in the original image can be analyzed and coded.

[0030] The content features of the original image include the cross-sectional size of the beam, the spacing of the column, the diameter and spacing of the reinforcement, etc. The feature code of the auxiliary watermark encodes these content features respectively, and each coding part corresponds to a specific content feature.

[0031] For example, for the content feature of the cross-sectional size of the beam, a corresponding coding part can be generated; for the content feature of the spacing of the column, a corresponding coding part will also be generated, and the various coding parts are combined in a set order to form a complete auxiliary watermark feature code.

[0032] Step S120: Domain feature analysis is performed on the original image to obtain the color feature domain and the texture feature domain of the original image. The color feature domain is composed of the pixel value distribution of each color channel, and the texture feature domain is composed of the texture structure patterns of different regions in the image.

[0033] The domain feature analysis of the original image is to separate different features of the image in order to subsequently embed the main watermark and the auxiliary watermark. First, the original image is analyzed from the aspects of color and texture to obtain the color feature domain and the texture feature domain.

[0034] The color feature domain emphasizes the distribution of colors in the image, while the texture feature domain emphasizes the texture structure characteristics of different regions in the image. Through the above domain feature analysis, the watermark embedding operation can be more targeted, improving the concealment and stability of the watermark.

[0035] Step S121: Channel separation is performed on the pixel color array of the original image, and the pixel color array is decomposed into a plurality of single-channel color sequences, each of which contains the color values of all pixel points in the channel.

[0036] When performing channel separation on the pixel color array of the original image, operation is performed according to the color model adopted by the image. If the image adopts the RGB color model, the pixel color array will be decomposed into three single-channel color sequences of red, green and blue.

[0037] During the decomposition process, the red component value of each pixel point is sequentially extracted from the pixel color array, and these red component values are arranged in order to form a red single-channel color sequence. Similarly, the green component value and the blue component value of each pixel point are extracted to form a green single-channel color sequence and a blue single-channel color sequence, respectively. Each single-channel color sequence contains the color values of all pixel points in the corresponding color channel, and the color information distribution of the channel is completely preserved.

[0038] Step S122: The plurality of single-channel color sequences are combined in the color channel order of the original image to generate a color feature domain, the arrangement order of each single-channel color sequence in the color feature domain is consistent with the color channel order of the original image, and the pixel value distribution of each color channel is composed of the color value range and the occurrence frequency in the corresponding single-channel color sequence.

[0039] After the separation of the single-channel color sequences is completed, the plurality of single-channel color sequences are combined according to the inherent order of the color channels of the original image. For example, for an image of the RGB color model, three single-channel color sequences are combined in the order of red, green and blue to form a color feature domain.

[0040] The arrangement order of each single-channel color sequence in the color feature domain strictly follows the color channel order of the original image to ensure the integrity and accuracy of the color information. The pixel value distribution of each color channel is embodied by the color value range and the occurrence frequency in the corresponding single-channel color sequence. The color value range refers to the interval between the minimum value and the maximum value of the color value in the channel, and the occurrence frequency refers to the proportion of the number of occurrences of different color values in the interval to the total number of pixel points.

[0041] Step S123: The spatial texture layout of the original image is regionally divided, and the spatial texture layout is divided into a plurality of texture regions according to the similarity of the texture direction and density, each of which is composed of a continuous pixel region.

[0042] When the spatial texture layout of the original image is regionally divided, the similarity of texture direction and density is taken as the basis for division. First, an initial division region is determined, the texture direction and density in the region are analyzed, and then the region is taken as the starting point for expansion.

[0043] In the expansion process, it is determined whether the texture direction and density of the adjacent region are similar to those of the initial region. If they are similar, they are included in the current division region. If they are not similar, they are determined as the starting point of a new division region, and the above expansion process is repeated.

[0044] In the above manner, the spatial texture layout is divided into multiple texture regions, each of which is composed of continuous pixel regions, and the texture direction and density in the region have high similarity.

[0045] Step S124: Extract the texture structure pattern of each texture region, which includes the direction, thickness and repetition rule of the texture.

[0046] When extracting the texture structure pattern of each texture region, the distribution of pixel points in each texture region is analyzed in detail. The direction of the texture is determined by calculating the gradient direction of the color value change of the pixel points in the region. The regions with consistent gradient direction have the same texture direction.

[0047] The thickness of the texture is determined by analyzing the number of changes of the texture structure in a unit length. The texture with fewer changes is relatively thick, and the texture with more changes is relatively thin. The repetition rule of the texture is determined by observing the occurrence of the texture structure at different positions. If a certain texture structure appears regularly in the region, the texture of the region has obvious repetition rule.

[0048] Through the above analysis, the texture structure pattern of each texture region is extracted, including the direction, thickness and repetition rule of the texture and other information.

[0049] Step S125: Arrange the texture structure patterns of all texture regions in the order of their positions in the spatial texture layout to generate a texture feature domain, which is composed of the texture structure patterns of different regions in the original image.

[0050] After extracting the texture structure patterns of all texture regions, the texture structure patterns are arranged in order according to the positions of these texture regions in the spatial texture layout. The arrangement order corresponds to the positions of the regions in space, starting from the top left corner of the image, and then arranging to the right and down until all texture regions are covered.

[0051] The texture feature domain generated by the above arrangement method completely contains the texture structure patterns of different regions in the original image, and can accurately reflect the distribution of the texture features of the image.

[0052] Step S130: embedding the main watermark into the color feature domain and embedding the auxiliary watermark into the texture feature domain to obtain the color feature domain with the main watermark and the texture feature domain with the auxiliary watermark.

[0053] After the color feature domain and the texture feature domain are extracted, the embedding operations of the main watermark and the auxiliary watermark are respectively performed. The main watermark is embedded into the color feature domain and the auxiliary watermark is embedded into the texture feature domain. In this way, the watermark can be closely combined with the features of the image, and the attack resistance of the watermark can be improved.

[0054] The embedding process needs to ensure the concealment of the watermark, that is, the visual effect of the original image will not be obviously affected after the watermark is embedded, and the stability of the watermark also needs to be ensured, that is, the watermark can still be accurately extracted after the image is processed to a certain extent.

[0055] Step S131: performing numerical conversion on the identification sequence of the main watermark, converting each character in the identification sequence into a numerical value that is adapted to the color value range of each single-channel color sequence in the color feature domain, and forming a main watermark numerical string after the numerical conversion, wherein each numerical value in the main watermark numerical string corresponds to a character in the identification sequence.

[0056] The numerical conversion on the identification sequence of the main watermark is to make the main watermark be able to be combined with the color values in the color feature domain. First, each character in the identification sequence is analyzed to determine the ASCII code value corresponding to each character.

[0057] Then, the color value range of each single-channel color sequence in the color feature domain, that is, the minimum and maximum values of the color values in each single-channel color sequence, is analyzed. According to these color value ranges, a conversion rule is determined to convert the ASCII code value of each character into a numerical value falling within the color value range of the corresponding single-channel color sequence.

[0058] The converted numerical values are arranged in the order of the characters in the identification sequence to form a main watermark numerical string, and each numerical value in the main watermark numerical string corresponds to a character in the identification sequence, which ensures the integrity of the main watermark information.

[0059] Step S1311: parsing the identification sequence of the main watermark, and extracting each character in the identification sequence to determine the character encoding value of each character.

[0060] When the identification sequence of the main watermark is parsed, the characters in the identification sequence are extracted one by one according to the arrangement order of the identification sequence. Each character has its corresponding character encoding value, for example, in the ASCII encoding standard, each character corresponds to a specific integer encoding value.

[0061] The character code value of each extracted character can be determined through a character code lookup table.

[0062] Step S1312: Analyze the color values of each single-channel color sequence in the color feature field to determine the minimum color value and the maximum color value of each single-channel color sequence, and the range between the two is the color value range of the single-channel color sequence.

[0063] When analyzing the color values of each single-channel color sequence in the color feature field, all color values in each single-channel color sequence are traversed to find the minimum value and the maximum value.

[0064] The interval between the minimum color value and the maximum color value is the color value range of the single-channel color sequence, which reflects the variation range of the color value of the channel.

[0065] Step S1313: Calculate the interval length of the color value range of each single-channel color sequence, which is the difference between the maximum color value and the minimum color value.

[0066] After determining the minimum color value and the maximum color value of each single-channel color sequence, the difference between the two is calculated, which is the interval length of the color value range of the single-channel color sequence.

[0067] The interval length reflects the size of the variation range of the color value of the channel. The larger the interval length, the richer the color variation of the channel.

[0068] Step S1314: According to the interval length and the character code value, convert the character code value of each character into a value falling within the color value range of the corresponding single-channel color sequence through a preset conversion algorithm, which ensures that the converted value does not exceed the color value range.

[0069] The preset conversion algorithm converts the value according to the interval length and the character code value. First, determine the range of the character code value, i.e. the minimum value and the maximum value among all character code values. Then calculate the proportion of the interval length of the character code value range to the interval length of the color value range of the single-channel color sequence.

[0070] For the encoding value of each character, first calculate the difference between it and the minimum value of the character code value, then multiply the difference by the above proportion to get an intermediate value, and finally add the minimum value of the color value of the single-channel color sequence to the intermediate value to get the converted value. Through the above conversion algorithm, it can be ensured that the converted value falls within the color value range of the corresponding single-channel color sequence.

[0071] Step S1315: Arrange the converted values in the order of the characters in the identification sequence to generate the main watermark value string, and each value in the main watermark value string corresponds to a character in the identification sequence one by one.

[0072] Each value obtained by the conversion algorithm is arranged in the original order of the characters in the main watermark identification sequence. The converted value of the first character is placed at the beginning of the main watermark value string, the converted value of the second character is placed at the second position, and so on.

[0073] The generated main watermark value string has a one-to-one correspondence between each value and the characters in the identification sequence, and the information of the main watermark is completely preserved.

[0074] Step S132: Traverse the plurality of single-channel color sequences in the color feature domain, and determine the pixel point positions to be modified according to a set selection rule, wherein the selection rule is constructed based on the distribution density of the pixel points in the single-channel color sequence.

[0075] When traversing the plurality of single-channel color sequences in the color feature domain, the arrangement order of the single-channel color sequences is followed. The selection rule is constructed based on the distribution density of the pixel points in the single-channel color sequence. In the region with a relatively high distribution density, i.e., the region where the pixel points are relatively concentrated, a relatively large number of pixel points to be modified are selected; in the region with a relatively low distribution density, a relatively small number of pixel points to be modified are selected.

[0076] The pixel point positions to be modified determined by the above selection rule can not only ensure the strength of watermark embedding, but also avoid excessive influence on the visual effect of the image. The pixel point positions to be modified are uniformly distributed in each single-channel color sequence, so as to improve the stability of the watermark.

[0077] Step S133: Replace the color values of the corresponding pixel points to be modified with the values in the main watermark value string in sequence to obtain the single-channel color sequence containing the main watermark.

[0078] The values in the main watermark value string are sequentially taken out and replaced with the color values of the pixel points to be modified one by one. The first value replaces the color value of the first pixel point to be modified, the second value replaces the color value of the second pixel point to be modified, and so on.

[0079] During the replacement process, the values in the main watermark value string directly overwrite the original color values of the pixel points to be modified. After the replacement is completed, each single-channel color sequence becomes a single-channel color sequence containing the main watermark, and the main watermark information is successfully embedded in the color sequence.

[0080] Step S134: Combine the single-channel color sequences containing the main watermark in the original color channel order to generate a color feature domain containing the main watermark, and the arrangement order of the single-channel color sequences in the color feature domain containing the main watermark is consistent with the arrangement order in the color feature domain.

[0081] After the watermark embedding of the single-channel color sequence is completed, the single-channel color sequences containing the main watermark are combined together in the order of the original color channels. For example, for an image of the RGB color model, the single-channel color sequences containing the main watermark are still combined in the order of red, green, and blue.

[0082] The arrangement order of each single-channel color sequence of the combined color feature field containing the main watermark remains consistent with the original color feature field, ensuring the structural integrity of the color information and also making the main watermark stably exist in the color feature field.

[0083] Step S135: Perform structural conversion on the feature encoding of the auxiliary watermark, convert the feature encoding into encoding symbols matching the texture structure patterns of each texture region in the texture feature field, and the converted encoding symbols form an auxiliary watermark encoding string, each encoding symbol in the auxiliary watermark encoding string corresponding to part of the feature encoding.

[0084] The structural conversion of the feature encoding of the auxiliary watermark is to make the auxiliary watermark be able to be integrated with the texture structure patterns in the texture feature field. First, the feature encoding of the auxiliary watermark is disassembled and decomposed into multiple independent feature units, each of which corresponds to a specific content feature in the original image.

[0085] Then, the texture structure patterns of each texture region in the texture feature field are analyzed to understand the description methods and feature expressions of different texture structure patterns. According to the characteristics of these texture structure patterns, a coding symbol system matching them is designed, and each coding symbol can find a corresponding position and expression form in the description of the texture structure pattern.

[0086] Each feature unit is converted into a corresponding encoding symbol according to the coding symbol system, and the converted encoding symbols are arranged in the order of the feature units in the feature encoding to form an auxiliary watermark encoding string, which ensures that the auxiliary watermark information can be accurately embedded into the texture feature field.

[0087] Step S1351: Disassemble the feature encoding of the auxiliary watermark to obtain multiple feature units, each of which corresponds to a specific content feature in the image.

[0088] When disassembling the feature encoding of the auxiliary watermark, the feature encoding is split according to its internal logical structure. The feature encoding is usually composed of multiple parts with specific meanings, and each part is distinguished by a specific delimiter.

[0089] By identifying these separators, the feature code is split into multiple independent feature units. For example, if the feature code is "L...W...F...", where "L...", "W..." and "F..." are separated by specific separators, it can be split into feature units corresponding to length, width and flatness respectively. Each feature unit corresponds to a specific content feature in the image, and the information of the feature code is completely retained.

[0090] Step S1352: Perform pattern classification on the texture structure patterns of the texture regions in the texture feature field, and determine the class identifier of each texture structure pattern, which is used to distinguish different types of texture structures.

[0091] When performing pattern classification on the texture structure patterns of the texture regions in the texture feature field, the direction, thickness and repetition law of the texture are used as the basis for operation. Texture structure patterns with the same or similar direction, thickness and repetition law are classified into the same class, and each class of texture structure pattern is assigned a unique class identifier.

[0092] The class identifier can be a letter, a number or a combination thereof, as long as it can effectively distinguish different types of texture structures. For example, texture structure patterns with horizontal direction, relatively thick and no obvious repetition law are marked as "A1", texture structure patterns with vertical direction, relatively thin and periodic repetition law are marked as "B2", and so on.

[0093] Step S1353: Establish a mapping relationship table between the feature units and the class identifiers, and each class identifier in the mapping relationship table corresponds to a value range of a feature unit.

[0094] When establishing the mapping relationship table, the value characteristics of each feature unit and the characteristics of the texture structure pattern represented by each class identifier are analyzed. According to the value range of the feature unit and the adaptation degree of the texture structure pattern, the value range of the feature unit corresponding to each class identifier is determined.

[0095] For example, if a feature unit represents the spacing of steel bars, its value range is within a set interval, and a certain class identifier corresponds to a texture structure pattern that matches the spacing range of the steel bar texture, then the corresponding relationship between the class identifier and the value range of the feature unit is recorded in the mapping relationship table. The mapping relationship table ensures that the feature unit can be accurately converted into an encoding symbol suitable for a specific texture structure pattern.

[0096] Step S1354: According to the mapping relationship table, convert each feature unit into the corresponding encoding symbol matched by the class identifier, and the format of the encoding symbol is consistent with the description format of the texture structure pattern.

[0097] According to the mapping relationship table, for each feature unit, its value range is determined first, and then the corresponding category identifier is found in the mapping relationship table. According to the found category identifier, the encoding symbol consistent with the description format of the texture structure mode represented by the category identifier is selected.

[0098] The format of the encoding symbol needs to match the description format of the texture structure mode, for example, if the description format of the texture structure mode contains a specific expression way of parameters such as direction and thickness, the encoding symbol also adopts a similar expression way, so as to ensure that the encoding symbol can be naturally embedded in the description of the texture structure mode and is not easily detected.

[0099] Step S1355: The converted encoding symbols are arranged in the order of the feature units in the feature code to generate an auxiliary watermark encoding string, and each encoding symbol in the auxiliary watermark encoding string corresponds to a feature unit in the feature code.

[0100] The encoding symbols converted from each feature unit are arranged in the original order of the feature units in the feature code. The encoding symbol corresponding to the first feature unit is placed at the first position of the auxiliary watermark encoding string, the encoding symbol corresponding to the second feature unit is placed at the second position, and so on.

[0101] In the generated auxiliary watermark encoding string, each encoding symbol corresponds to a feature unit in the feature code, and the information of the auxiliary watermark is completely preserved.

[0102] Step S136: A plurality of texture regions in the texture feature domain are traversed, and according to the type of the texture structure mode of each texture region, the encoding symbol in the auxiliary watermark encoding string is embedded in the description of the corresponding texture structure mode to obtain a texture region containing an auxiliary watermark.

[0103] When traversing the plurality of texture regions in the texture feature domain, the arrangement order of the texture regions in the texture feature domain is followed. For each texture region, the type of the texture structure mode, i.e., the corresponding category identifier, is determined first.

[0104] Then, the encoding symbol matched with the category identifier is taken out from the auxiliary watermark encoding string, and the encoding symbol is embedded in the description of the texture structure mode of the texture region. The embedding position is selected in a part of the description of the texture structure mode that does not affect the expression of its main features, so as to ensure that the overall features of the texture structure mode do not change significantly after the encoding symbol is embedded.

[0105] After the encoding symbol embedding of one texture region is completed, the next texture region is processed, and all encoding symbols are embedded in the corresponding texture regions to obtain a plurality of texture regions containing an auxiliary watermark.

[0106] Step S137: arranging the auxiliary-watermark-containing texture regions in the original region position order to generate an auxiliary-watermark-containing texture feature domain, and the arrangement order of each texture region in the auxiliary-watermark-containing texture feature domain is consistent with the arrangement order in the texture feature domain.

[0107] All the auxiliary-watermark-containing texture regions are arranged in the position order in the original texture feature domain. Starting from the first texture region of the texture feature domain, the auxiliary-watermark-containing texture regions are sequentially arranged, and the arrangement order is kept the same as that in the original texture feature domain.

[0108] The auxiliary-watermark-containing texture feature domain generated after the arrangement not only retains the structural integrity of the original texture feature domain, but also stably exists in the texture feature domain, and is not easy to be removed or tampered.

[0109] Step S140: performing watermark extraction and comparison on the main-watermark-containing color feature domain and the auxiliary-watermark-containing texture feature domain to generate a double verification result, and the double verification result is used to indicate the legality of the source and the integrity of the content of the image.

[0110] After the watermark embedding is completed, the main-watermark-containing color feature domain and the auxiliary-watermark-containing texture feature domain need to be subjected to watermark extraction and comparison to verify the security of the image. The extraction process corresponds to the embedding process, and it is ensured that the embedded main watermark and auxiliary watermark can be accurately extracted.

[0111] By comparing the extracted watermark with the preset watermark, it is judged whether the source of the image is legal and the content is complete, and a double verification result is generated.

[0112] Step S141: extracting the main watermark from the main-watermark-containing color feature domain to obtain the extracted main watermark, and the extraction process is to determine the pixel point position according to the selection rule in the embedding process, extract the color value of the pixel point and convert it into the corresponding character, and then combine them in order to form the extracted main watermark.

[0113] When extracting the main watermark from the main-watermark-containing color feature domain, the operation process is opposite to that in the embedding process. By determining the pixel point position modified in the embedding process, the color value is extracted and converted, and finally combined into the extracted main watermark to verify whether the main watermark exists completely.

[0114] Step S1411: obtaining a plurality of single-channel color sequences in the main-watermark-containing color feature domain, and the arrangement order of each single-channel color sequence is consistent with the arrangement order of the single-channel color sequence in the color feature domain.

[0115] When the multiple single-channel color sequences in the color feature domain containing the main watermark are acquired, the single-channel color sequences are extracted according to the arrangement order of the single-channel color sequences in the color feature domain containing the main watermark. The arrangement order of the single-channel color sequences is consistent with the arrangement order of the single-channel color sequences in the original color feature domain, so that the position of the embedded main watermark can be accurately found.

[0116] Step S1412: According to the selection rule set when the main watermark is embedded, the positions of the modified pixel points in each single-channel color sequence are determined, and the positions of the modified pixel points are the same as the positions of the pixel points to be modified when the main watermark is embedded.

[0117] According to the selection rule set when the main watermark is embedded, the positions of the modified pixel points in each single-channel color sequence are determined. Since the selection rule is fixed, the positions of the modified pixel points are completely the same as the positions of the pixel points to be modified when the main watermark is embedded, so that the position of the main watermark information can be accurately found.

[0118] Step S1413: The color values of the pixel points are extracted, each color value is converted into a corresponding character code value through an inverse algorithm of the numerical conversion, and then the character code value is converted into a corresponding character.

[0119] After the color values of the determined pixel points are extracted, each color value is processed by using the inverse algorithm of the numerical conversion. The inverse algorithm first subtracts the minimum value of the color value of the corresponding single-channel color sequence from the color value to obtain a difference value, then divides the difference value by the ratio of the range interval length of the character code value to the range interval length of the color value of the single-channel color sequence to obtain another difference value, and finally adds the minimum value of the character code value to the difference value to obtain the corresponding character code value.

[0120] The obtained character code value is converted into a corresponding character through a character code query table to restore the character information of the main watermark.

[0121] Step S1414: The converted characters are arranged according to the position order of the pixel points in the single-channel color sequence to generate the extracted main watermark, and the identification sequence length of the extracted main watermark is the same as the preset identification sequence length of the main watermark.

[0122] The converted characters are arranged according to the position order of the pixel points in the single-channel color sequence. The characters corresponding to the pixel points appearing first in the single-channel color sequence are placed in front, and the characters corresponding to the pixel points appearing later are placed behind.

[0123] After the arrangement is completed, the generated extracted main watermark has the same identification sequence length as the preset identification sequence length of the main watermark, so that effective comparison can be performed.

[0124] Step S142: Extract the auxiliary watermark from the texture feature field containing the auxiliary watermark, to obtain the extracted auxiliary watermark. The extraction process is as follows: according to the category of the texture structure pattern of the texture region during embedding, extract the code symbol from the texture structure pattern description, convert the code symbol into the corresponding feature unit, and then combine the feature units in sequence to obtain the extracted auxiliary watermark.

[0125] When extracting the auxiliary watermark from the texture feature field containing the auxiliary watermark, the operation process is opposite to that during embedding. By determining the category of the texture structure pattern of the texture region, the code symbol is extracted and converted into a feature unit, and finally combined into the extracted auxiliary watermark to verify whether the auxiliary watermark exists completely.

[0126] Step S1421: Obtain a plurality of texture regions in the texture feature field containing the auxiliary watermark, and the arrangement order of each texture region is consistent with the arrangement order of the texture region in the texture feature field.

[0127] When obtaining a plurality of texture regions in the texture feature field containing the auxiliary watermark, the extraction is performed according to the arrangement order of the texture region in the texture feature field containing the auxiliary watermark. The arrangement order of the texture region is consistent with the arrangement order of the texture region in the original texture feature field, which ensures that the position of the embedded auxiliary watermark can be accurately found.

[0128] Step S1422: According to the category of the texture structure pattern during embedding, extract the embedded code symbol from the texture structure pattern description of each texture region. The format of the extracted code symbol is consistent with the format of the code symbol generated during the structure conversion.

[0129] According to the category of the texture structure pattern of each texture region during embedding, the corresponding texture structure pattern description is searched to find the embedded code symbol. The format of the extracted code symbol is completely consistent with the format of the code symbol generated during the structure conversion, which ensures that the feature unit can be accurately converted.

[0130] Step S1423: Convert each code symbol into the corresponding feature unit through the inverse mapping relationship during the structure conversion. The attribute of the feature unit is consistent with the attribute of the feature unit obtained during the decomposition of the feature code of the auxiliary watermark.

[0131] Each extracted code symbol is processed according to the inverse mapping relationship during the structure conversion. The inverse mapping relationship finds the corresponding category identifier according to the code symbol, and then determines the value range of the corresponding feature unit according to the category identifier, and then converts it into the corresponding feature unit.

[0132] The attribute of the converted feature unit, such as the content feature type represented, is completely consistent with the attribute of the feature unit obtained during the decomposition of the feature code of the auxiliary watermark, which ensures the accuracy of the auxiliary watermark information.

[0133] Step S1424: arranging the converted feature units in the order of the positions of the texture regions in the texture feature field to generate the extracted secondary watermark, the feature code length of the extracted secondary watermark being the same as the preset feature code length of the secondary watermark.

[0134] The converted feature units are arranged in the order of the positions of the texture regions in the texture feature field. The feature units corresponding to the texture regions processed first are placed in front, and the feature units corresponding to the texture regions processed later are placed behind.

[0135] The generated extracted secondary watermark has the same feature code length as the preset feature code length of the secondary watermark.

[0136] Step S143: comparing the extracted primary watermark with the preset primary watermark to check whether the identification sequence of the extracted primary watermark is the same as that of the preset primary watermark, obtaining a primary watermark comparison result, the primary watermark comparison result being same or different.

[0137] The identification sequence of the extracted primary watermark is compared with the identification sequence of the preset primary watermark character by character. Starting from the first character, the characters at each position are compared in sequence to determine whether they are consistent. If all the characters are completely the same, the primary watermark comparison result is same. If there is any character that is different, the primary watermark comparison result is different. The primary watermark comparison result can reflect whether the source information of the image is tampered with.

[0138] Step S144: comparing the extracted secondary watermark with the preset secondary watermark to check whether the feature code of the extracted secondary watermark is the same as that of the preset secondary watermark, obtaining a secondary watermark comparison result, the secondary watermark comparison result being same or different.

[0139] The feature code of the extracted secondary watermark is compared with the feature code of the preset secondary watermark feature unit by feature unit. Starting from the first feature unit, the contents of each feature unit are compared in sequence to determine whether they are consistent. If all the feature units are completely the same, the secondary watermark comparison result is same. If there is any feature unit that is different, the secondary watermark comparison result is different. The secondary watermark comparison result can reflect whether the content information of the image is tampered with.

[0140] Step S145: generating a double verification result according to the primary watermark comparison result and the secondary watermark comparison result. When the primary watermark comparison result and the secondary watermark comparison result are both same, the double verification result is that the image source is legal and the content is complete. When the primary watermark comparison result is different or the secondary watermark comparison result is different, the double verification result is that the image source is illegal or the content is tampered with.

[0141] The double verification result is generated by combining the main watermark comparison result and the auxiliary watermark comparison result. When both are the same, it indicates that the source information of the image is not tampered and the content information is complete, and the double verification result is that the image source is legal and the content is complete. When any one of the comparison results is different, it indicates that the image source information is tampered or the content information is tampered, and the double verification result is that the image source is illegal or the content is tampered.

[0142] The double verification result comprehensively reflects the security state of the image, and is a key basis for triggering the image security protection mechanism.

[0143] Step S150: Triggering the image security protection mechanism based on the double verification result, the image security protection mechanism including allowing the image to be normally used or starting an abnormal interception process.

[0144] According to the generated double verification result, the corresponding image security protection mechanism is started. If the image source is legal and the content is complete, the image is allowed to be normally used; if the image source is illegal or the content is tampered, the abnormal interception process is started to prevent illegal use of the image and take corresponding processing measures to ensure the security of the image.

[0145] Step S151: When the double verification result is that the image source is legal and the content is complete, a use permission signal is generated, and the use permission signal is used to authorize various operations on the original image, including viewing, copying and transmitting.

[0146] When the double verification result is that the image source is legal and the content is complete, the system automatically generates a use permission signal. The use permission signal contains authorized operation type information, such as viewing, copying, transmitting, etc., and clearly defines the legal operation range that can be performed on the original image.

[0147] The use permission signal is generated by a specific encoding method to ensure that it cannot be easily tampered with during transmission and can accurately convey the authorization information.

[0148] Step S152: The use permission signal is sent to the image use control component, so that after the image use control component receives the use permission signal, the use restriction on the original image is removed and the user is allowed to perform various legal operations.

[0149] The generated use permission signal is sent to the image use control component through an internal data transmission channel. After the image use control component receives the use permission signal, it is decoded and verified to confirm the legality and validity of the signal.

[0150] After the verification, the image usage control component removes all usage restrictions on the original image, and the user can perform legal operations such as viewing, copying, and transmitting the original image according to the operation type authorized by the usage permission signal, ensuring that the image is normally used under the premise of safety.

[0151] Step S153: When the double verification result is that the image source is illegal or the content is tampered, an abnormal interception signal is generated, which is used to trigger a target protection operation, including prohibiting image usage, recording abnormal information, and notifying the management system.

[0152] When the double verification result is that the image source is illegal or the content is tampered, the system immediately generates an abnormal interception signal. The abnormal interception signal contains information such as the type of exception, the time of occurrence, and is used to trigger a pre-set target protection operation, such as prohibiting image usage, recording abnormal information, and notifying the management system.

[0153] The generation of the abnormal interception signal is real-time, which can be issued at the first time of discovering the image exception, and timely prevent the spread and use of abnormal images.

[0154] Step S154: Send the abnormal interception signal to the image security interception component, so that the image security interception component receives the abnormal interception signal and terminates any operation request on the original image, records the type, time, and source information of the image exception and sends it to the pre-set management system.

[0155] The abnormal interception signal is sent to the image security interception component through an internal emergency data transmission channel. After receiving the abnormal interception signal, the image security interception component immediately starts the interception program, terminates any ongoing or imminent operation request on the original image, and prevents the abnormal image from being further processed or used.

[0156] At the same time, the image security interception component records the type of image exception in detail, such as illegal source or content tampering; records the exact time of the abnormal occurrence; records the source information of the image, such as the transmission path, the sender identifier, etc. After sorting these recorded information, it is sent to the pre-set management system through a special communication line, so that the management personnel can timely understand the situation and take corresponding measures.

[0157] Step S155: After the image security protection mechanism is executed, a protection execution report is generated, which contains the type of protection mechanism triggered, the execution time, and the execution result.

[0158] Regardless of whether the image security protection mechanism allows normal use of the image or initiates an abnormal interception process, after its execution is completed, the system will automatically generate a protection execution report. The protection execution report records in detail the type of protection mechanism triggered, i.e. normal use permission or abnormal interception; records the specific time of protection mechanism execution; records the execution result, such as which operations the user successfully performed, or the specific circumstances of abnormal interception, etc. The protection execution report comprehensively reflects the execution process and final effect of this security protection mechanism.

[0159] Step S156: Store the protection execution report into a designated record database, which is used to save the security protection history information of all original images.

[0160] After generating the protection execution report, it is transmitted to the designated record database through a data storage interface. The record database adopts a distributed storage architecture, has high reliability and high security, and can ensure that the protection execution report will not be lost or tampered with.

[0161] The record database saves the security protection history information of all original images, including each protection execution report, corresponding image identifier, processing time, etc. These information are stored in chronological order and image identifier for classification, which is convenient for subsequent query, statistics and analysis.

[0162] Through the long-term accumulation of security protection history information, the management personnel can master the running status of the image security protection mechanism and discover potential security risks. At the same time, these historical information can also be used as evidence when needed, to trace the use and security processing process of the image.

[0163] The method further includes a model training step for optimizing the accuracy of watermark embedding and extraction, and the specific steps are as follows: Step S210: Collect a large number of image samples in the field of construction engineering, which cover different types of building components, different shooting environments and different resolution images, and collect the main watermark and auxiliary watermark information corresponding to these samples.

[0164] When collecting image samples, ensure the diversity and representativeness of the samples. The building components covered include beams, columns, plates, walls and other types, the shooting environments include sunny, cloudy, indoor, outdoor and other different lighting and scene conditions, and the resolution includes multiple common specifications.

[0165] For each image sample, the corresponding main watermark and auxiliary watermark are generated in advance. The identification sequence of the main watermark contains the source information of the sample, and the feature code of the auxiliary watermark contains the content feature information of the sample. These samples and corresponding watermark information constitute the basic data set for model training.

[0166] Step S220: Preprocess the collected image samples, including image enhancement, noise removal, and size standardization operations, to improve the stability and accuracy of model training.

[0167] The image enhancement operation adjusts the contrast, brightness, and other parameters of the image to make the features more obvious, facilitating subsequent feature extraction and watermark processing. The noise removal operation uses a filtering algorithm to eliminate random noise and interference signals in the image, ensuring image quality.

[0168] The size standardization operation adjusts all image samples to a uniform size, avoiding the impact of image size differences on model training. The preprocessed image samples have more consistent feature representations while maintaining the original key information.

[0169] Step S230: Construct a neural network model optimized for watermark embedding and extraction, which includes an input layer, multiple hidden layers, and an output layer. The hidden layers include convolutional layers, pooling layers, and fully connected layers.

[0170] The input layer receives preprocessed image samples and corresponding watermark information, converting them into tensor form suitable for neural network processing. The convolutional layer uses multiple convolution kernels to perform convolution operations on the input tensor, extracting local features of the image and features of the watermark.

[0171] The pooling layer performs down-sampling on the feature maps output by the convolutional layer, reducing the dimensionality of the feature data while preserving key features. The fully connected layer integrates and maps the features output by the pooling layer, implementing the fusion of image features and watermark features.

[0172] The output layer outputs optimized watermark embedding parameters and extraction parameters, which guide the actual watermark embedding and extraction process, improving the accuracy and efficiency of watermark processing.

[0173] Step S240: Divide the preprocessed image samples and corresponding watermark information into training set, validation set, and test set. The training set is used for model parameter learning, the validation set is used to adjust the model's hyperparameters, and the test set is used to evaluate the model's final performance.

[0174] The division process uses random sampling to ensure consistent sample distribution in the three data sets. The training set has a large number of samples, providing sufficient learning data for the model; the validation set is used to monitor the model's performance during training, allowing for timely adjustments to learning rate, number of hidden layers, and other hyperparameters; the test set is used to objectively evaluate the model's generalization ability and processing effect after training is complete.

[0175] Step S250: Set the parameters of model training, including learning rate, iteration times, batch size, etc., and use cross-entropy loss function to calculate the error between the predicted value and the true value of the model.

[0176] The learning rate determines the speed of updating the model parameters, and setting a suitable learning rate can make the model converge to the optimal solution faster; the iteration times refer to the number of complete training of the model on the training set, and sufficient iteration times can ensure that the model learns the sample features sufficiently; the batch size refers to the number of samples used for each model parameter update, and a reasonable batch size can balance the training efficiency and training effect.

[0177] The cross-entropy loss function can effectively measure the difference between the optimized parameters of the model output and the true optimal parameters, and through the calculation of the loss value, it guides the model to adjust the parameters.

[0178] Step S260: Train the neural network model using the training set, in each iteration process, input the image sample into the model to get the predicted watermark embedding parameters and extraction parameters, calculate the error through the loss function, and update the weight parameters of the model using the back propagation algorithm.

[0179] During the training process, the model continuously receives image samples and corresponding watermark information, and calculates the predicted watermark embedding parameters and extraction parameters through forward propagation. Compare the predicted parameters with the true parameters, and calculate the error value through the cross-entropy loss function.

[0180] According to the error value, the back propagation algorithm is used to adjust the weight parameters of the model layer by layer from the output layer, so that the error value is continuously reduced. This process is repeated until the loss value of the model reaches the preset threshold or the set iteration times are completed.

[0181] Step S270: During the model training process, periodically validate the model using the validation set, and adjust the hyperparameters of the model according to the validation result. When the loss value of the validation set no longer decreases, stop the model training to avoid overfitting of the model.

[0182] After completing the set number of iterations, input the validation set into the model in training to calculate the loss value of the validation set. According to the change of the loss value of the validation set, adjust the hyperparameters of the model, such as increasing the learning rate to speed up convergence, or reducing the number of hidden layers to avoid overfitting.

[0183] When the loss value of the validation set no longer decreases for several consecutive iterations, it means that the model has reached the optimal state, at which point the training is stopped to prevent the model from overlearning the detailed features of the training set and reducing the generalization ability to new samples.

[0184] Step S280: The trained neural network model is evaluated for performance using a test set, and the accuracy, bit error rate, and other indicators of watermark embedding and extraction are calculated. If the evaluation results meet the preset requirements, the model training is complete and can be put into use. If not, the model structure and training parameters are adjusted, and the training is re-performed.

[0185] The image samples and watermark information of the test set do not participate in the training and validation process of the model, and can objectively reflect the actual performance of the model. During evaluation, the test set is input into the model to obtain the results of watermark embedding and extraction, which are compared with the true results to calculate the accuracy and bit error rate.

[0186] The accuracy refers to the proportion of the number of samples correctly embedded and extracted to the total number of samples, and the bit error rate refers to the proportion of the number of incorrect characters or feature units in the watermark information to the total number. If these indicators reach the preset threshold, it means that the model performance meets the requirements and can be applied to the actual watermark processing process; otherwise, the reasons need to be analyzed, the structure of the model needs to be adjusted, such as increasing the number of hidden layers or changing the size of the convolution kernel, or adjusting the training parameters, such as learning rate and iteration number, and re-training until the model performance meets the requirements.

[0187] Through the above model training steps, the optimized neural network model can more accurately determine the position and method of watermark embedding, as well as the parameters of watermark extraction, improve the reliability and efficiency of the entire multi-domain watermark double verification method, and better adapt to the needs of image security protection in the field of construction engineering.

[0188] Figure 2 A schematic diagram of exemplary hardware and software components of the multi-domain watermark double verification based image security protection system 100 that can implement the idea of the present application is shown. For example, the processor 120 can be used in the multi-domain watermark double verification based image security protection system 100 and used to perform the functions in the present application.

[0189] The multi-domain watermark double verification based image security protection system 100 can be a general-purpose server or a special-purpose server, both of which can be used to implement the multi-domain watermark double verification based image security protection method of the present application. Although only one server is shown in the present application, for the sake of convenience, the functions described in the present application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0190] For example, the multi-domain watermark double verification based image security protection system 100 can include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and different forms of storage media 140, such as a disk, a ROM, or a RAM, or any combination thereof. Illustratively, the multi-domain watermark double verification based image security protection system 100 can also include program instructions stored in a ROM, a RAM, or other types of non-transitory storage media, or any combination thereof. The methods of the present application can be implemented according to these program instructions. The multi-domain watermark double verification based image security protection system 100 also includes an I / O interface 150 between the computer and other input / output devices.

[0191] For ease of illustration, only one processor is described in the multi-domain watermark double verification based image security protection system 100. However, it should be noted that the multi-domain watermark double verification based image security protection system 100 in the present application can also include multiple processors, so the steps performed by one processor described in the present application can also be jointly performed or separately performed by multiple processors. For example, if the processor of the multi-domain watermark double verification based image security protection system 100 performs steps A and B, it should be understood that steps A and B can also be jointly performed by two different processors or separately performed in one processor. For example, a first processor performs step A, a second processor performs step B, or the first processor and the second processor jointly perform steps A and B.

[0192] In addition, the present application also provides a readable storage medium, wherein computer executable instructions are pre-set in the readable storage medium, and when a processor executes the computer executable instructions, the multi-domain watermark double verification based image security protection method is implemented.

[0193] It should be noted that, in order to simplify the description of the present application and to help understand one or more embodiments of the present application, in the foregoing description of the embodiments of the present application, various features are sometimes combined into one embodiment, drawing, or description thereof.

Claims

1. An image security protection method based on multi-domain watermark double verification, characterized in that, The method comprises: obtaining an original image and preset primary and secondary watermarks, the original image containing a pixel color array and a spatial texture layout, the primary watermark being an identification sequence associated with the source of the image, and the secondary watermark being a feature code corresponding to the content of the image; performing domain feature analysis on the original image to obtain a color feature domain and a texture feature domain of the original image, the color feature domain being composed of pixel value distribution of each color channel, and the texture feature domain being composed of texture structure patterns of different regions in the image; embedding the primary watermark into the color feature domain and the secondary watermark into the texture feature domain to obtain a color feature domain containing the primary watermark and a texture feature domain containing the secondary watermark; performing watermark extraction and comparison on the color feature domain containing the primary watermark and the texture feature domain containing the secondary watermark to generate a double verification result; triggering an image security protection mechanism based on the double verification result, the image security protection mechanism including allowing normal use of the image or starting an abnormal interception process.

2. The image security protection method based on multi-domain watermark double verification according to claim 1, characterized in that, The method comprises: receiving an externally transmitted original image, the original image existing in the form of a two-dimensional pixel matrix, each element in the two-dimensional pixel matrix representing color information of a pixel point; extracting a pixel color array from the two-dimensional pixel matrix of the original image, the pixel color array being composed of color values of pixel points arranged in matrix order; analyzing the distribution rule of pixel points in the original image to determine the texture direction and density of different regions and generate a spatial texture layout, the texture feature of each region in the spatial texture layout being determined by the arrangement mode of continuous pixels; calling a primary watermark from a preset watermark library, the identification sequence of the primary watermark being arranged by specific characters according to a preset rule, and the identification sequence being associated with the source information of the original image; calling a secondary watermark from the preset watermark library, the feature code of the secondary watermark being generated based on the content features of the original image, and each part of the feature code corresponding to a specific content feature in the original image. 3.The image security protection method based on multi-domain watermark double verification according to claim 1, characterized in that, The method comprises: performing channel separation on the pixel color array of the original image to decompose the pixel color array into a plurality of single-channel color sequences, each single-channel color sequence containing color values of all pixel points in the channel; combining the plurality of single-channel color sequences according to the color channel order of the original image to generate a color feature domain, the arrangement order of each single-channel color sequence in the color feature domain being consistent with the color channel order of the original image, and the pixel value distribution of each color channel being composed of the color value range and frequency of occurrence in the corresponding single-channel color sequence; performing region division on the spatial texture layout of the original image to divide the spatial texture layout into a plurality of texture regions according to the similarity of texture direction and density, each texture region being composed of a continuous pixel region; extracting a texture structure pattern of each texture region, the texture structure pattern including the direction, thickness and repetition rule of the texture; Arranging the texture structure patterns of all texture regions in the order of the positions of the regions in the spatial texture layout to generate a texture feature domain, the texture feature domain being composed of the texture structure patterns of different regions in the original image.

4. The image security protection method based on multi-domain watermark double verification according to claim 1, characterized in that, The embedding the main watermark into the color feature domain and the embedding the auxiliary watermark into the texture feature domain to obtain a color feature domain containing the main watermark and a texture feature domain containing the auxiliary watermark, comprises: Converting the identification sequence of the main watermark into numerical values, converting each character in the identification sequence into a numerical value that is adapted to the color value range of each single-channel color sequence in the color feature domain, the converted numerical values forming a main watermark numerical string, each numerical value in the main watermark numerical string corresponding to a character in the identification sequence one by one; Traversing the multiple single-channel color sequences in the color feature domain, determining the pixel point positions to be modified according to a set selection rule, wherein the selection rule is constructed based on the distribution density of the pixel points in the single-channel color sequences; Replacing the color values of the corresponding pixel points to be modified with the numerical values in the main watermark numerical string in sequence to obtain single-channel color sequences containing the main watermark; Combining the single-channel color sequences containing the main watermark in the original color channel order to generate a color feature domain containing the main watermark, the arrangement order of the single-channel color sequences in the color feature domain containing the main watermark being consistent with the arrangement order in the color feature domain; Converting the feature code of the auxiliary watermark into encoding symbols that are matched with the texture structure patterns of the texture regions in the texture feature domain, the converted encoding symbols forming an auxiliary watermark encoding string, each encoding symbol in the auxiliary watermark encoding string corresponding to part of the feature code one by one; Traversing the multiple texture regions in the texture feature domain, embedding the encoding symbols in the auxiliary watermark encoding string into the corresponding texture structure pattern descriptions according to the type of the texture structure pattern of each texture region to obtain texture regions containing the auxiliary watermark; Arranging the texture regions containing the auxiliary watermark in the original region position order to generate a texture feature domain containing the auxiliary watermark, the arrangement order of the texture regions in the texture feature domain containing the auxiliary watermark being consistent with the arrangement order in the texture feature domain.

5. The image security protection method based on multi-domain watermark double verification according to claim 4, characterized in that, The converting the identification sequence of the main watermark into numerical values, converting each character in the identification sequence into a numerical value that is adapted to the color value range of each single-channel color sequence in the color feature domain, comprises: Parsing the identification sequence of the main watermark, extracting each character in the identification sequence, and determining the character encoding value of each character; Analyzing the color values of each single-channel color sequence in the color feature domain, determining the minimum color value and the maximum color value of each single-channel color sequence, the range between the two being the color value range of the single-channel color sequence; Calculating the interval length of the color value range of each single-channel color sequence, the interval length being the difference between the maximum color value and the minimum color value; According to the interval length and the character encoding value, converting the character encoding value of each character into a numerical value that falls within the color value range of the corresponding single-channel color sequence through a preset conversion algorithm, the conversion algorithm ensuring that the converted numerical value does not exceed the color value range; The converted values are arranged in the order of the characters in the identification sequence to generate a primary watermark value string, each value in the primary watermark value string corresponding to a character in the identification sequence.

6. The image security protection method based on multi-domain watermark double verification according to claim 4, characterized in that, The feature code of the secondary watermark is structurally converted to convert the feature code into code symbols matching the texture structure patterns of the texture regions in the texture feature domain, and the converted code symbols form a secondary watermark code string, including: The feature code of the secondary watermark is decomposed to obtain a plurality of feature units, each feature unit corresponding to a specific content feature in the image; The texture structure patterns of the texture regions in the texture feature domain are classified to determine the class identifiers of each texture structure pattern, the class identifiers being used to distinguish different types of texture structures; A mapping relationship table between the feature units and the class identifiers is established, each class identifier in the mapping relationship table corresponding to a value range of a feature unit; According to the mapping relationship table, each feature unit is converted into a code symbol matched with the corresponding class identifier, the format of the code symbol being consistent with the description format of the texture structure pattern; The converted code symbols are arranged in the order of the feature units in the feature code to generate a secondary watermark code string, each code symbol in the secondary watermark code string corresponding to a feature unit in the feature code.

7. The image security protection method based on multi-domain watermark double verification according to claim 1, characterized in that, The color feature domain containing the primary watermark and the texture feature domain containing the secondary watermark are subjected to watermark extraction and comparison to generate a dual verification result, including: The primary watermark is extracted from the color feature domain containing the primary watermark to obtain an extracted primary watermark, the extraction process being to determine the pixel point positions according to the selection rules during embedding, extract the color values of the pixel points and convert them into corresponding characters, and then combine the characters in sequence to form the extracted primary watermark; The secondary watermark is extracted from the texture feature domain containing the secondary watermark to obtain an extracted secondary watermark, the extraction process being to extract the code symbols from the texture structure pattern description according to the class of the texture structure pattern of the texture region during embedding, convert the code symbols into corresponding feature units, and then combine the feature units in sequence to form the extracted secondary watermark; The extracted primary watermark is compared with the preset primary watermark to check whether the identification sequences of the extracted primary watermark and the preset primary watermark are completely identical, and a primary watermark comparison result is obtained, the primary watermark comparison result being identical or different; The extracted secondary watermark is compared with the preset secondary watermark to check whether the feature codes of the extracted secondary watermark and the preset secondary watermark are completely identical, and a secondary watermark comparison result is obtained, the secondary watermark comparison result being identical or different; The dual verification result is generated according to the primary watermark comparison result and the secondary watermark comparison result, when the primary watermark comparison result and the secondary watermark comparison result are both identical, the dual verification result being that the image source is legal and the content is complete; when the primary watermark comparison result is different or the secondary watermark comparison result is different, the dual verification result being that the image source is illegal or the content is tampered.

8. The image security protection method based on multi-domain watermark double verification according to claim 7, characterized in that, The primary watermark is extracted from the color feature domain containing the primary watermark to obtain an extracted primary watermark, including: A plurality of single-channel color sequences in the color feature domain containing the primary watermark are obtained, the arrangement order of each single-channel color sequence being consistent with the arrangement order of the single-channel color sequences in the color feature domain; According to the selection rule set during embedding, the pixel position modified in each single-channel color sequence is determined, which is the same as the pixel position to be modified during embedding; Extract the color value of the pixel, convert each color value to the corresponding character code value through the inverse algorithm during numerical conversion, and then convert the character code value to the corresponding character; Arrange the converted characters in the order of the pixel position in the single-channel color sequence to generate the extracted main watermark, and the length of the identification sequence of the extracted main watermark is the same as the length of the identification sequence of the preset main watermark; In addition, the auxiliary watermark is extracted from the texture feature domain containing the auxiliary watermark to obtain the extracted auxiliary watermark, including: Obtain a plurality of texture regions in the texture feature domain containing the auxiliary watermark, and the arrangement order of each texture region is consistent with the arrangement order of the texture regions in the texture feature domain; According to the category identifier of the texture structure pattern during embedding, extract the embedded code symbol in the texture structure pattern description of each texture region, and the format of the extracted code symbol is consistent with the format of the code symbol generated during structure conversion; Convert each code symbol to the corresponding feature unit through the inverse mapping relationship during structure conversion, and the attribute of the feature unit is consistent with the attribute of the feature unit obtained during the feature encoding of the auxiliary watermark; Arrange the converted feature units in the order of the position of the texture region in the texture feature domain to generate the extracted auxiliary watermark, and the feature encoding length of the extracted auxiliary watermark is the same as the feature encoding length of the preset auxiliary watermark.

9. The image security protection method based on multi-domain watermark double verification according to claim 1, characterized in that, The image security protection mechanism is triggered based on the double verification result, including: When the double verification result is that the image source is legal and the content is complete, a use permission signal is generated, which is used to authorize various operations on the original image, including viewing, copying and transmitting; Send the use permission signal to the image use control component, so that the image use control component receives the use permission signal, removes the use restriction on the original image and allows the user to perform various legal operations; When the double verification result is that the image source is illegal or the content is tampered, an abnormal interception signal is generated, which is used to trigger the target protection operation, including prohibiting image use, recording abnormal information and notifying the management system; Send the abnormal interception signal to the image security interception component, so that the image security interception component receives the abnormal interception signal, terminates any operation request on the original image, records the type, time and source information of the image exception and sends them to the preset management system; After the execution of the image security protection mechanism is completed, a protection execution report is generated, which contains the type of triggered protection mechanism, execution time and execution result; Store the protection execution report in the designated record database, which is used to save the security protection history information of all original images.

10. An image security protection system based on multi-domain watermark double verification, characterized in that, The application relates to an image security protection method based on multi-domain watermark double verification, comprising a processor and a memory, the memory and the processor being connected, the memory being used for storing programs, instructions or codes, and the processor being used for executing the programs, instructions or codes in the memory to realize the image security protection method based on multi-domain watermark double verification according to any one of claims 1-9.

Images