A watermark embedding method, a watermark identification method, and an electronic device

Abstract

The application provides a watermark embedding method, a watermark identification method and an electronic device. The watermark embedding method comprises: processing a carrier image to obtain first low-frequency features, generating s first sub-rings based on the first low-frequency features; processing a watermark image to obtain second low-frequency features, generating s second sub-rings based on the second low-frequency features; performing encryption processing on the s second sub-rings to obtain s encrypted second sub-rings; determining s zero watermarks based on the s first sub-rings and the corresponding encrypted second sub-rings; and embedding the s zero watermarks in the carrier image to obtain a carrier image containing watermark information. The application performs ring-type encryption on the carrier image and the copyright watermark, can improve the security of the watermark information, and can realize multi-copyright sharing.

Description

Technical Field

[0001] This application relates to the field of watermark processing technology, and in particular to a watermark embedding method, a watermark recognition method, and an electronic device. Background Technology

[0002] Existing blind watermarking algorithms embed watermark information into the low-frequency domain or singular values ​​of an image. They can achieve good resistance to single-type attacks, meaning they are robust to geometric attacks but poor to non-geometric attacks, or they exhibit good resistance to non-geometric attacks but cannot extract clear copyright images for geometric attacks.

[0003] Currently, blind watermarking methods still have the following shortcomings:

[0004] Most existing blind watermarking methods use traditional frequency domain transformation methods to embed watermark information into the low frequency and maximum singular value of the carrier image. They can achieve good robustness against non-geometric or geometric attacks, but they are still difficult to extract clear copyright images under combined attacks.

[0005] Most existing watermarking methods are based on block-based embedding of watermark information. When the image is subjected to geometric deformation, the synchronization of the watermark will be severely affected.

[0006] Existing blind watermarking does not consider the correction of watermark information errors after an attack, which severely affects the clarity of the extracted watermark image. Moreover, the encryption method is relatively simple, which is not conducive to information security and cannot achieve copyright sharing. Summary of the Invention

[0007] In view of this, this application provides a watermark embedding method, a watermark recognition method, and an electronic device to solve the above-mentioned technical problems.

[0008] In a first aspect, embodiments of this application provide a watermark embedding method, including:

[0009] The carrier image is processed to obtain the first low-frequency feature, and s first sub-rings are generated based on the first low-frequency feature;

[0010] The watermark image is processed to obtain the second low-frequency feature, and s second sub-rings are generated based on the second low-frequency feature;

[0011] Encrypting s second sub-rings yields s encrypted second sub-rings;

[0012] Based on s first sub-rings and their corresponding encrypted second sub-rings, determine s zero watermarks;

[0013] Embedding s zero watermarks into the carrier image yields a carrier image containing watermark information.

[0014] In some possible implementations, the carrier image is processed to obtain a first low-frequency feature, and s first sub-rings are generated based on the first low-frequency feature, including:

[0015] The carrier image is converted to the YCbCr color space to obtain the Y component;

[0016] Performing a dual-tree complex wavelet transform on the Y component yields the first low-frequency feature. ,in, This represents the series of the decomposition of the dual-tree complex wavelet transform. Representing frequency domain subbands in different directions;

[0017] Divide the first low-frequency feature into s first sub-rings: ,in, For the i-th first sub-ring, .

[0018] In some possible implementations, s second sub-rings are encrypted to obtain s encrypted second sub-rings; including:

[0019] Randomly generate s positioning matrices The i-th positioning matrix Representing the i-th key ;

[0020] Using the i-th key For the i-th second sub-ring Perform encryption to obtain the encrypted i-th second sub-ring. :

[0021]

[0022] in, This is an XOR operation.

[0023] In some possible implementations, based on s first sub-rings and their corresponding encrypted second sub-rings, s zero watermarks are determined, including:

[0024] For the i-th first sub-ring Performing an improved Schur decomposition yields the i-th upper triangular matrix. ;

[0025] For the i-th upper triangular matrix Perform a binarization operation to obtain the i-th binary feature matrix. :

[0026]

[0027] in, This is the modulo operation function;

[0028] Based on the encrypted i-th second sub-ring With the i-th binary characteristic matrix Get the i-th zero watermark :

[0029]

[0030] in, This is an XOR operation.

[0031] In some possible implementations, the zero watermark is embedded in the carrier image to obtain a carrier image containing watermark information, including:

[0032] Based on the i-th zero watermark For the upper triangular matrix of the i-th first sub-ring first row and first column element After processing, the upper triangular matrix with the i-th embedded watermark is obtained. Maximum Singularity :

[0033]

[0034] in, To quantize the step size, For embedding strength; The first remainder: ;

[0035] For the i-th upper triangular matrix with embedded watermark Reconstruct the feature matrix to obtain the i-th feature matrix containing the watermark information. :

[0036]

[0037] in, It is a unitary matrix. for The transpose of the matrix;

[0038] For the i-th feature matrix containing watermark information Reconstruct the i-th third sub-ring containing the watermark information;

[0039] Perform a dual-tree complex wavelet inverse transform on the i-th third sub-ring containing watermark information to obtain the i-th fourth sub-ring;

[0040] Image reconstruction is performed on all fourth sub-rings to obtain a carrier image containing watermark information.

[0041] Secondly, embodiments of this application provide a watermark recognition method, including:

[0042] The carrier image containing watermark information is processed to obtain the third low-frequency feature, and s fifth sub-rings are generated based on the third low-frequency feature;

[0043] An improved Schur decomposition is performed on each fifth sub-ring to obtain s upper triangular matrices;

[0044] Based on s upper triangular matrices, determine s first watermarks;

[0045] Using s keys, the corresponding first watermark is decrypted to obtain s second watermarks, where the second watermark is the decrypted first watermark;

[0046] The pre-trained optimized network is used to process the s second watermarks to obtain s third watermarks.

[0047] In one possible implementation, based on s upper triangular matrices, s first watermarks are determined, including:

[0048] Based on the i-th upper triangular matrix Determine the i-th first watermark :

[0049]

[0050] in, To quantize the step size, For embedding strength: The second remainder: ; For matrix The element in the first row and first column.

[0051] In one possible implementation, the optimized network comprises, in sequence, downsampling units, a first convolutional layer, a first batch processing layer, a first ReLU function, a second convolutional layer, a second batch processing layer, a second ReLU function, a third convolutional layer, a third batch processing layer, a third ReLU function, and a fully connected layer; wherein the kernel size of the first convolutional layer is [missing information]. The kernel size of the second convolutional layer is The kernel size of the third convolutional layer is ;

[0052] The method further includes:

[0053] Establish a training set, which includes: multiple sets of watermarked image samples, each set of watermarked image samples including the original watermarked image and the watermarked image after the attack;

[0054] An optimized watermark image is obtained by processing the attacked watermark image using an optimized network.

[0055] The loss value is determined based on the optimized watermarked image and the original watermarked image;

[0056] The parameters of the network are updated and optimized based on the loss value.

[0057] Thirdly, embodiments of this application provide an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the watermark embedding method or watermark recognition method of embodiments of this application.

[0058] Fourthly, an embodiment of this application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the watermark embedding method or watermark recognition method of the embodiments of this application.

[0059] This application employs ring encryption on the carrier image and copyright watermark, which enhances the security of the watermark information and enables multiple copyright sharing. Attached Figure Description

[0060] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0061] Figure 1 A schematic diagram of the division of the annular sub-bands provided in an embodiment of this application;

[0062] Figure 2 A flowchart of the watermark embedding method provided in the embodiments of this application;

[0063] Figure 3 A flowchart of the watermark recognition method provided in the embodiments of this application;

[0064] Figure 4 A schematic diagram of the peak signal-to-noise ratio of a carrier image after embedding an encrypted watermark, provided in an embodiment of this application.

[0065] Figure 5 The normalized correlation coefficient value of the carrier image provided in the embodiments of this application after a cropping attack;

[0066] Figure 6 The normalized correlation coefficient values ​​of the carrier image provided in this application embodiment under attacks of different median filter sizes;

[0067] Figure 7The changes in normalized correlation coefficient values ​​obtained after JPEG compression attacks of different intensities provided in the embodiments of this application;

[0068] Figure 8 A functional structure diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0069] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0070] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0071] First, a brief introduction to the design concept of the embodiments of this application will be given.

[0072] To overcome the technical problems of current digital watermarking, this application provides a dual watermarking method that integrates ring embedding and deep learning. A dual-tree complex wavelet transform (DT-CWT) is performed on the carrier image to obtain a stable embedding domain, eliminating the influence of random signals on the watermark. Ring embedding units are then constructed in the stable low-frequency domain to improve the synchronization of the watermark under rotation attacks and multiple attacks. Secondly, an improved Schur decomposition is used for the constructed ring region to further enhance the stability of the watermark. Simultaneously, to balance the robustness and invisibility of the watermark, deep learning is used to train the watermark embedding strength to reduce the visual impact of the watermark on the image. The system also trains the watermark image after an attack and generates a copyright image highly similar to the original watermark. In the event of a copyright dispute, the trained network model can improve the clarity of the watermark. Using different encryption methods to encrypt different watermark rings for the same watermark and embedding the encrypted watermark information in the encryption domain can improve information security.

[0073] The inventive points of this application include:

[0074] 1. Ring encryption

[0075] Existing blind watermarking technologies rely on relatively simple encryption methods, resulting in low security and hindering copyright sharing. Most methods employ a block-based approach, preprocessing and encrypting the image and watermark separately to enhance security. However, this invention utilizes a ring-shaped method to generate watermark sub-rings. and image sub-ring :

[0076]

[0077] in, For carrier image, This is the original copyright watermark. To further enhance the unpredictability of the copyright watermark, a location matrix containing multiple representation keys is randomly generated. , Then, by judging the positioning matrix The elements in the code are used to select different encryption methods to encrypt the watermark sub-ring:

[0078]

[0079] Because the copyright of some digital works is sometimes not owned by an individual or a single organization, but jointly by multiple individuals or organizations, a work with shared copyright can only be held by one person, making it impossible for multiple parties to share the copyright, thus causing internal copyright disputes among members. To achieve key sharing and facilitate subsequent copyright information management, a Lagrange interpolation method is used to distribute keys.

[0080] The user set is The shared key set is Each element in the shared key set satisfies ,and elements in The elements in the text correspond one-to-one. Used to solve the coefficient terms in the key-sharing equation key The decryption formula is used to solve for the chaotic mapping parameters or scrambling times of the encryption algorithm:

[0081]

[0082] in, The initial value for the encryption method, To use the corresponding key The obtained chaotic mapping parameters or number of scrambling iterations. When the user... Upon obtaining the key, first use and initial value Substitute the constant term into the decryption formula Then the key Substitute the parameters into the decryption formula to obtain the chaotic mapping parameters.

[0083] 2. Ring-based watermark embedding strategy

[0084] The strategy consists of two stages: ring partitioning and zero-watermark generation. In the first stage, to reduce the impact of geometric attacks on watermark synchronization, ring sub-bands are partitioned before watermark embedding. First, the low-frequency domain of the carrier image is divided into different ring-shaped sub-bands centered on the image center, such as... Figure 1 As shown, the distance between all pixels and the center of the circle is calculated using the following formula. :

[0085]

[0086] Divide all pixels that are equidistant into a ring. The distance is the radius of the annulus, and this calculation continues until the last annulus with the smallest radius is reached.

[0087] in, Let the coordinates be the center of the circle. These are the coordinates of the image pixels. This is a rounding function. Where is the radius of the annulus. Let be the number of concentric circles, and satisfy the following conditions: . Figure 1 It displays concentric circles with different radii, where the numbers on the yellow sub-blocks represent the distance from the corresponding pixel to the center of the circle, i.e., the radius of the ring.

[0088] The second stage is used to generate a zero watermark, after obtaining the embedded ring of the image. Then, firstly according to Divide each ring into two equal parts, dividing each part into the left and right halves. and the right half ring The DCT coefficients of the cyclic segments are extracted using Discrete Cosine Transform (DCT). Calculate the first 9 DCT coefficients to obtain the energy information for each half-ring. Secondly, based on the energy level of each half-ring, image features are generated. :

[0089]

[0090] Finally, the extracted image features are combined with the encrypted watermark to generate zero-watermark information, which is then used to embed subsequent watermark embeddings.

[0091] 3. Improved Schur decomposition

[0092] Schur decomposition, as a matrix factorization method for feature extraction, effectively removes image redundancy and decomposes into translation-invariant and robust features. It has wide applications in image recognition, information hiding, and digital forensics. For a given image of size... matrix There exists a unitary matrix , making

[0093]

[0094] in, It is a unitary matrix. It is an upper triangular matrix, because the matrix It contains a large amount of feature information of the matrix, therefore embedding the watermark information into the maximum singular value of the image... In order to resist conventional attacks, embedding watermark information into an image can be considered as noise signal. Most existing watermarking methods directly quantize and embed the watermark information into the image. In the first element of the matrix :

[0095]

[0096] in, The original maximum singular value, The maximum singular value after embedding watermark information. For embedding strength, This is watermark information. Without any attack, Represented as

[0097]

[0098] When a watermarked image is attacked, the default attack signal, like the watermark information, acts additively on the image's pixel values, affecting the watermark information. :

[0099]

[0100] in, This represents the maximum value of the watermark after an attack. When a watermarked image is attacked, the watermark extraction is...

[0101]

[0102] in, For the extracted watermark, The attack information shows that the clarity of the watermark information is directly affected by the strength of the attack signal. When the attack signal strength increases, the attack signal will directly cover the watermark information. To further reduce the impact of the attack signal on the watermark information, this embodiment improves the traditional Schur decomposition:

[0103]

[0104] in, For the modulation matrix, As a regulating factor, used to control The magnitude of the value, when the improved Schur decomposition decomposes the attacked image, can be determined by... This reduces the range of singular values.

[0105] 4. Deep learning models

[0106] Information transmission in the channel is affected by noise, compression, or filtering, which can affect the copyright information of watermarked images. The extracted watermarked images contain a lot of noise. In order to further improve the clarity of the watermarked images, an intermediate copyright image generation model is designed to optimize the extracted watermark.

[0107] Compared with the prior art, the beneficial effects of this application are as follows:

[0108] 1. Encrypting the carrier image and copyright watermark in a ring pattern can improve the security of watermark information and enable multiple copyright sharing.

[0109] 2. The ring-based embedding method is more robust than the block-based watermarking method, which can improve the robustness of the algorithm. Even if the watermarked image is subjected to geometric deformation, the feature information of the image sub-rings is less affected, thereby improving the algorithm's resistance to shearing attacks and row and column offsets. At the same time, the multidirectionality and singularity of DT-CWT can reduce the impact of noise attacks, JPEG compression and filtering attacks on the image.

[0110] 3. Based on the improved Schur decomposition, the resistance performance of watermarks is further enhanced, and the ring watermark embedding and extraction method effectively improves the capacity of embedded watermarks.

[0111] 4. Using deep learning methods to train the attacked watermark image can generate a copyright image with a very high similarity to the original watermark through the network, which helps to improve the clarity of the watermark.

[0112] After introducing the application scenarios and design concepts of the embodiments of this application, the technical solutions provided by the embodiments of this application will be described below.

[0113] like Figure 2 As shown, this application provides a watermark embedding method, including:

[0114] Step 101: Process the carrier image to obtain the first low-frequency feature, and generate s first sub-rings based on the first low-frequency feature;

[0115] Step 102: Process the watermark image to obtain the second low-frequency feature, and generate s second sub-rings based on the second low-frequency feature;

[0116] Step 103: Encrypt the s second sub-rings to obtain s encrypted second sub-rings;

[0117] Step 104: Based on the s first sub-rings and the corresponding encrypted second sub-rings, determine s zero watermarks;

[0118] Step 105: Embed s zero watermarks into the carrier image to obtain a carrier image containing watermark information.

[0119] This embodiment constructs circular sub-bands on both the carrier image and the watermark image, which can reduce the impact of geometric attacks on watermark synchronization and improve watermark security.

[0120] In some embodiments, the carrier image is processed to obtain a first low-frequency feature, and s first sub-rings are generated based on the first low-frequency feature, including:

[0121] The carrier image is converted to the YCbCr color space to obtain the Y component;

[0122] Performing a dual-tree complex wavelet transform on the Y component yields the first low-frequency feature. ,in, This represents the series of the decomposition of the dual-tree complex wavelet transform. Representing frequency domain subbands in different directions;

[0123] Divide the first low-frequency feature into s first sub-rings: ,in, For the i-th first sub-ring, .

[0124] In some embodiments, s second sub-rings are encrypted to obtain s encrypted second sub-rings; including:

[0125] Randomly generate s positioning matrices The i-th positioning matrix Representing the i-th key ;

[0126] Using the i-th key For the i-th second sub-ring Perform encryption to obtain the encrypted i-th second sub-ring. :

[0127]

[0128] in, This is an XOR operation.

[0129] This embodiment uses different keys to encrypt different sub-rings, which can improve the security of the watermark.

[0130] In some embodiments, determining s zero watermarks based on s first sub-rings and corresponding encrypted second sub-rings includes:

[0131] For the i-th first sub-ring Performing an improved Schur decomposition yields the i-th upper triangular matrix. ;

[0132] For the i-th upper triangular matrix Perform a binarization operation to obtain the i-th binary feature matrix. :

[0133]

[0134] in, This is the modulo operation function;

[0135] Based on the encrypted i-th second sub-ring With the i-th binary characteristic matrix Get the i-th zero watermark :

[0136]

[0137] in, This is an XOR operation.

[0138] This embodiment employs an improved Schur decomposition, which can refine image features and improve the accuracy and robustness of watermark information.

[0139] In some embodiments, embedding a zero watermark into a carrier image to obtain a carrier image containing watermark information includes:

[0140] Based on the i-th zero watermark For the upper triangular matrix of the i-th first sub-ring first row and first column element After processing, the upper triangular matrix with the i-th embedded watermark is obtained. Maximum Singularity :

[0141]

[0142] in, To quantize the step size, For embedding strength; The first remainder: ;

[0143] For the i-th upper triangular matrix with embedded watermark Reconstruct the feature matrix to obtain the i-th feature matrix containing the watermark information. :

[0144]

[0145] in, It is a unitary matrix. for The transpose of the matrix;

[0146] For the i-th feature matrix containing watermark information Reconstruct the i-th third sub-ring containing the watermark information;

[0147] Perform a dual-tree complex wavelet inverse transform on the i-th third sub-ring containing watermark information to obtain the i-th fourth sub-ring;

[0148] Image reconstruction is performed on all fourth sub-rings to obtain a carrier image containing watermark information.

[0149] like Figure 3 As shown in the figure, this application provides a watermark recognition method, including:

[0150] Step 201: Process the carrier image containing watermark information to obtain the third low-frequency feature, and generate s fifth sub-rings based on the third low-frequency feature;

[0151] Step 202: Perform improved Schur decomposition on each fifth sub-ring to obtain s upper triangular matrices;

[0152] Step 203: Determine s first watermarks based on s upper triangular matrices;

[0153] Step 204: Use s keys to decrypt the corresponding first watermarks to obtain s second watermarks, where the second watermarks are the decrypted first watermarks;

[0154] Step 205: Process the s second watermarks using the pre-trained optimized network to obtain s third watermarks.

[0155] In some embodiments, determining s first watermarks based on s upper triangular matrices includes:

[0156] Based on the i-th upper triangular matrix Determine the i-th first watermark :

[0157]

[0158] in, To quantize the step size, For embedding strength: The second remainder: ; For matrix The element in the first row and first column.

[0159] In some embodiments, the optimized network includes, in sequence, a downsampling unit, a first convolutional layer, a first batch processing layer, a first ReLU function, a second convolutional layer, a second batch processing layer, a second ReLU function, a third convolutional layer, a third batch processing layer, a third ReLU function, and a fully connected layer; wherein the kernel size of the first convolutional layer is [missing information]. The step size is 2. The kernel size of the second convolutional layer is The step size is 2. The kernel size of the third convolutional layer is The step size is 2. .

[0160] The method further includes:

[0161] A training set is established, which includes: multiple sets of watermarked image samples, each set of watermarked image samples including the original watermarked image and the attacked watermarked image; wherein the attack methods include geometric attacks, non-geometric attacks and a combination of both.

[0162] An optimized watermark image is obtained by processing the attacked watermark image using an optimized network.

[0163] Based on the optimized watermarked image and the original watermarked image, the loss value is determined using the minimum mean square error.

[0164] The parameters of the network are updated and optimized based on the loss value.

[0165] The specific implementation process of this application will be described below using a specific application scenario.

[0166] This embodiment uses Peak Signal-to-Noise Ratio (PSNR) to evaluate the visual effect of the embedded watermark image; a higher PSNR indicates better invisibility. Furthermore, Normalized Correlation (NC) is used to measure the similarity between the original watermark and the watermark extracted after the attack; a higher NC value indicates a clearer extracted watermark and better robustness. Experimental results are as follows: Figure 4 , Figure 5 , Figure 6 and Figure 7 As shown; where, Figure 4 PSNR values ​​after embedding watermarks on 6 different carrier images. Figure 5 The NC values ​​are for six different carrier images after a cropping attack. Figure 6 The NC values ​​are for six carrier images under attacks of different median filter sizes. Figure 7 The changes in NC values ​​obtained after JPEG compression attacks of different intensities.

[0167] Based on the above embodiments, this application also provides an electronic device, see below. Figure 8 As shown, the electronic device 300 provided in this application embodiment includes at least: a processor 301, a memory 302, and a computer program stored in the memory 302 and executable on the processor 301. When the processor 301 executes the computer program, it implements the watermark embedding method or watermark recognition method provided in this application embodiment.

[0168] The electronic device 300 provided in this application embodiment may further include a bus 303 connecting different components (including processor 301 and memory 302). The bus 303 represents one or more types of bus structures, including memory bus, peripheral bus, local area bus, etc.

[0169] The memory 302 may include a readable medium in the form of volatile memory, such as random access memory (RAM) 3021 and / or cache memory 3022, and may further include read-only memory (ROM) 3023.

[0170] The memory 302 may also include a program tool 3025 having a set (at least one) of program modules 3024, including but not limited to: an operating subsystem, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0171] Electronic device 300 can also communicate with one or more external devices 304 (e.g., keyboard, remote control, etc.), and with one or more devices that enable a user to interact with electronic device 300 (e.g., mobile phone, computer, etc.), and / or with any device that enables electronic device 300 to communicate with one or more other electronic devices 300 (e.g., router, modem, etc.). This communication can be performed through input / output (I / O) interface 305. Furthermore, electronic device 300 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) through network adapter 306. Figure 8 As shown, network adapter 306 communicates with other modules of electronic device 300 via bus 303. It should be understood that, although... Figure 8 As not shown, other hardware and / or software modules may be used in conjunction with electronic device 300, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, Redundant Arrays of Independent Disks (RAID) subsystems, tape drives, and data backup storage subsystems.

[0172] It should be noted that, Figure 8 The electronic device 300 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0173] This application also provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the watermark embedding method or watermark recognition method provided in this application. Specifically, the executable program can be built into or installed in the electronic device 300, so that the electronic device 300 can implement the watermark embedding method or watermark recognition method provided in this application by executing the built-in or installed executable program.

[0174] The watermark embedding method or watermark recognition method provided in this application embodiment can also be implemented as a program product. The program product includes program code. When the program product can run on the electronic device 300, the program code is used to make the electronic device 300 execute the watermark embedding method or watermark recognition method provided in this application embodiment.

[0175] The program product provided in this application embodiment can be any combination of one or more readable media, wherein the readable media can be a readable signal medium or a readable storage medium, and the readable storage medium can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination thereof. Specifically, more specific examples of readable storage media (a non-exhaustive list) include: electrical connections with one or more wires, portable disks, hard disks, RAM, ROM, erasable programmable read-only memory (EPROM), optical fibers, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0176] The program product provided in this application embodiment can be a CD-ROM and include program code, and can also run on a computing device. However, the program product provided in this application embodiment is not limited thereto. In this application embodiment, the readable storage medium can be any tangible medium that contains or stores a program, which can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0177] It should be noted that although several units or sub-units of the device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of this application, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided and embodied by multiple units.

[0178] Furthermore, although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0179] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and are not intended to limit it. Although this application has been described in detail with reference to the embodiments, those skilled in the art should understand that modifications or equivalent substitutions to the technical solutions of this application do not depart from the spirit and scope of the technical solutions of this application, and should all be covered within the scope of the claims of this application.

Claims

1. A watermark embedding method, characterized in that, include: The carrier image is processed to obtain the first low-frequency feature, and s first sub-rings are generated based on the first low-frequency feature; The watermark image is processed to obtain the second low-frequency feature, and s second sub-rings are generated based on the second low-frequency feature; Encrypting s second sub-rings yields s encrypted second sub-rings; Based on s first sub-rings and their corresponding encrypted second sub-rings, determine s zero watermarks; Embed s zero watermarks into the carrier image to obtain a carrier image containing watermark information; The carrier image is processed to obtain a first low-frequency feature, and s first sub-rings are generated based on the first low-frequency feature, including: The carrier image is converted to the YCbCr color space to obtain the Y component; Performing a dual-tree complex wavelet transform on the Y component yields the first low-frequency feature. ,in, This represents the series of the decomposition of the dual-tree complex wavelet transform. Representing frequency domain subbands in different directions; Divide the first low-frequency feature into s first sub-rings: ,in, For the i-th first sub-ring, ; Encrypting s second sub-rings yields s encrypted second sub-rings; including: Randomly generate s positioning matrices The i-th positioning matrix Representing the i-th key ; Using the i-th key For the i-th second sub-ring Perform encryption to obtain the encrypted i-th second sub-ring. : in, This is an XOR operation; Based on s first sub-rings and their corresponding encrypted second sub-rings, determine s zero watermarks, including: For the i-th first sub-ring Performing an improved Schur decomposition yields the i-th upper triangular matrix. ; The improved Schur decomposition is as follows: in, For the modulation matrix, As a regulating factor, It is a unitary matrix; for the i-th upper triangular matrix Perform a binarization operation to obtain the i-th binary feature matrix. : in, This is the modulo operation function; Based on the encrypted i-th second sub-ring With the i-th binary characteristic matrix Get the i-th zero watermark : 。 2. The watermark embedding method according to claim 1, characterized in that, Embedding the zero watermark into the carrier image yields a carrier image containing watermark information, including: Based on the i-th zero watermark For the upper triangular matrix of the i-th first sub-ring first row and first column element After processing, the upper triangular matrix with the i-th embedded watermark is obtained. Maximum Singularity : : : in, To quantize the step size, For embedding strength; The first remainder: For the i-th upper triangular matrix with embedded watermark Reconstruct the feature matrix to obtain the i-th feature matrix containing the watermark information. : in, It is a unitary matrix. for The transpose of the matrix; For the i-th feature matrix containing watermark information Reconstruct the i-th third sub-ring containing the watermark information; Perform a dual-tree complex wavelet inverse transform on the i-th third sub-ring containing watermark information to obtain the i-th fourth sub-ring; Image reconstruction is performed on all fourth sub-rings to obtain a carrier image containing watermark information.

3. A watermark recognition method, characterized in that, include: The carrier image containing watermark information is processed to obtain the third low-frequency feature, and s fifth sub-rings are generated based on the third low-frequency feature; A modified Schur decomposition is performed on each fifth sub-ring to obtain s upper triangular matrices: The improved Schur decomposition is as follows: in, For the modulation matrix, As a regulating factor, Given a unitary matrix; based on s upper triangular matrices, determine s first watermarks; Using s keys, the corresponding first watermark is decrypted to obtain s second watermarks, where the second watermark is the decrypted first watermark; The pre-trained optimized network is used to process the s second watermarks to obtain s third watermarks; The optimized network comprises, in sequence, downsampling units, a first convolutional layer, a first batch processing layer, a first ReLU function, a second convolutional layer, a second batch processing layer, a second ReLU function, a third convolutional layer, a third batch processing layer, a third ReLU function, and a fully connected layer; wherein, the kernel size of the first convolutional layer is [missing information]. The kernel size of the second convolutional layer is The kernel size of the third convolutional layer is ; The method further includes: Establish a training set, which includes: multiple sets of watermarked image samples, each set of watermarked image samples including the original watermarked image and the watermarked image after the attack; An optimized watermark image is obtained by processing the attacked watermark image using an optimized network. The loss value is determined based on the optimized watermarked image and the original watermarked image; The parameters of the network are updated and optimized based on the loss value.

4. The watermark recognition method according to claim 3, characterized in that, Based on s upper triangular matrices, determine s first watermarks, including: Based on the i-th upper triangular matrix Determine the i-th first watermark : in, To quantize the step size, For embedding strength: The second remainder: For matrix The element in the first row and first column.

5. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as claimed in any one of claims 1-2 or the method as claimed in any one of claims 3-4.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the method as claimed in any one of claims 1-2 or the method as claimed in any one of claims 3-4.

Images