33 results about "Watermark algorithm" patented technology

The invention relates to the field of watermarking algorithms, and provides a watermarking method and system for generating a text by a large language model. The method comprises the steps that a prompt text and a generated text sequence are input into a generation model, and an original logic score of a next mark position is generated through probability distribution calculation; performing semantic feature extraction on the generated text sequence by using an embedding model to obtain semantic embedding representation; converting the semantic embedding representation into a watermark logic score based on a pre-trained watermark model; and performing weighted combination on the original logic score and the watermark logic score to generate a final logic score, and outputting a watermark mark based on the final logic score. According to the method, the high robustness of the watermark is improved, and the safety of the watermark is also improved.

The invention relates to an alliance chain-based energy data sharing method and device, and a medium, and the method comprises the steps: collecting and preprocessing energy data based on a TEE, encrypting the data in the TEE, uploading the data to an IPFS, generating a hash value, and binding a data owner identifier; when the verification request is responded, integrity verification, numerical value verification and other operations are executed, and an encryption result is generated and returned to the demand side. After verification is passed, copyright information is embedded according to a data type matching watermarking algorithm, the IPFS is re-encrypted and uploaded, and a shared hash value is obtained. And storing the hash value and the decryption key into the block chain through the consensus node of the alliance chain, and transmitting the hash value and the decryption key to the demand side. And after the demand side decrypts and obtains the data, extracting the watermark and comparing the watermark with the on-chain parameters for verification. According to the method, sharing and verification of the data in an encryption state are ensured, privacy protection and use controllability are considered, and the problems of data leakage, tampering and ownership tracing in traditional sharing are solved.

The invention discloses a medical image robust watermarking algorithm based on a re-parameterized high-efficiency attention network, and relates to the field of image data processing methods, and the algorithm comprises the steps: obtaining an original medical image; a pre-trained RepEA-Net model is used for extracting multi-scale deep features with good generalization performance from an original medical image, and then a mean perceptual hash algorithm is used for generating feature vectors. Obtaining an original watermark image; the watermark image is encrypted by using a Logistic chaotic encryption algorithm, the security of the watermark image is improved, and finally, the encrypted watermark image and the binary feature vector are subjected to XOR operation to generate the zero watermark. Through the innovative design of the re-parameterization efficient attention network, the contradiction among robustness, generalization and image integrity of a traditional medical image watermarking algorithm is broken through, and a solution with technical advancement and clinical practicability is provided for copyright protection and safe transmission of medical images in the medical Internet of Things.

The application provides a kind of based on Transform and prompt guide imperceptible watermark attack method, belongs to digital information security technical field.Method includes: through existing data set and QPHFMs watermark algorithm construction training data set, adjusts the size of original watermark image and original non-watermark image containing uniformly, obtains standardization training data;Imperceptible watermark attack network TFP-WAN containing Transform encoder-decoder framework and dynamic prompt module is constructed, dynamic prompt module includes PGM and PIM;MAE is used as loss function in TFP-WAN;Standardization training data is input into TFP-WAN and iteratively trained until reaching preset iteration number, and the watermark attack model of training completion is obtained;The image after attack is output after inputting the watermark image to be attacked into watermark attack model.The application realizes the attack to robust watermark algorithm with higher error rate.

This invention discloses a deep learning-based Hadamard transform-based anti-screen-capture watermarking method, belonging to the field of watermarking technology. This invention combines a convolutional neural network (CNN) and residual blocks to achieve an end-to-end process of watermark embedding and extraction within the Hadamard domain. By simulating screen-capture attacks and incorporating this method between the embedding and extraction layers, the network ensures robust watermark embedding. This method can efficiently extract watermark information from photos taken by surreptitiously, protecting the copyright of digital media. The use of the transform domain for watermark embedding allows the watermark to spread over a wider area of ​​the image, significantly improving the robustness of the watermarking algorithm. Comparison with related technologies demonstrates the superiority of this invention in terms of imperceptibility and robustness.

The invention discloses a watermark embedding method, and the method comprises the steps: determining a first watermark field used for embedding watermark information based on a first data set; determining m groups of first embedded data rows based on the first data set; wherein m is an integer greater than or equal to 2; obtaining first attribute values corresponding to the first watermark fields in the m groups of first embedded data rows; based on each group of first attribute values, determining a second attribute value corresponding to the first watermark field in each group of first embedded data rows; wherein each group of second attribute values is used for indicating a used watermark algorithm; modifying each group of first attribute values into the corresponding second attribute values to obtain a second data set; and embedding the watermark information into the attribute value corresponding to the first watermark field in the second data set by adopting the watermark algorithm to obtain a third data set. The invention further discloses a watermark extraction method and device, equipment, a storage medium and a computer program product.

The embodiment of the present application discloses a text traceability detection method and system, which can effectively improve the amount of embedded information and robustness of the watermark algorithm. By using invisible characters to convert the watermark signal into corresponding codes, a large amount of watermark information is hidden and embedded; by embedding the watermark image into the document carrier, the robust restoration of the ownership identification is achieved. The copyright tracing of the document and the detection of the source of leakage can also be achieved at the same time. The user's user identification image is encrypted and decrypted using the personal key of the copyright user to complete the carrier copyright tracing of the document; the extracted watermark identification is repaired by the DNA base repair algorithm to detect the source of leakage of the document.

The invention discloses an image watermarking algorithm based on Kyber fully homomorphic encryption, which is characterized in that an anti-quantum Kyber lattice cryptographic algorithm is combined with a Patchwork watermarking algorithm, so that the exchangeability (first encryption and then embedding or first embedding and then encryption) of an encryption and watermark embedding sequence is realized, and dual-mode watermark extraction of a ciphertext domain and a plaintext domain is supported. According to the method, the addition homomorphic characteristic of Kyber is utilized, it is ensured that the operation sequence does not affect ciphertext generation and watermark extraction, the safety defect of a traditional algorithm in the quantum environment is overcome, and the method is suitable for scenes of digital copyright management, medical data sharing and the like and has the advantages of quantum attack resistance, high robustness, outstanding privacy protection capacity and the like.

The invention discloses a data privacy tracking method for a power system in a data-in-channel environment. The data privacy tracking method comprises the following steps: acquiring data information of a target power system; establishing a blood relationship graph of data in the target power system; constructing a single shot relationship between the blood relationship path and the watermark parameter; executing a watermark algorithm, and embedding watermark information; and data privacy tracking of the target power system in the data medium environment is realized. The invention also discloses a system for realizing the data privacy tracking method used in the data-in-channel environment of the power system. According to the method, through the construction of the blood relationship graph and the construction of the single shot relationship, the data privacy tracking in the data in-channel environment of the power system is realized, and the method is higher in reliability, better in accuracy and higher in applicability.

This invention discloses a method and apparatus for evaluating the robustness of image watermarking. The method includes: acquiring an original image dataset; printing watermarks on a portion of the original images in the original image dataset using the watermarking algorithm to be evaluated, forming a watermarked image dataset; attacking all images in the watermarked image dataset to obtain an attacked image dataset; performing watermark recognition on all images in the attacked image dataset to obtain watermark recognition results; the watermark recognition results include the true positive rate and the false positive rate; using a preset image quality measurement algorithm, calculating the image quality measurement values ​​before and after the attack for each watermarked image, as the image quality degradation result; determining the image watermark robustness evaluation result based on the watermark recognition result and the image quality degradation result; the image watermark robustness evaluation result includes an image watermark robustness evaluation report. This invention provides a complete, systematic, and mature method for evaluating watermarking algorithms.

The invention discloses a color image blind watermarking method based on two-dimensional quantization index modulation, and the method comprises the steps: firstly carrying out the discrete Fourier transform of image blocks, and converting a pixel matrix into a frequency domain coefficient matrix; then, energy distribution is further optimized through Laplacian transformation, so that energy is obviously gathered in a left diagonal region and a right diagonal region of the matrix, and two maximum energy coefficients are obtained; on the basis, a two-dimensional quantization index modulation technology is innovatively utilized, and the robustness of a watermark algorithm is enhanced by optimizing a quantization strategy. The method not only has better invisibility, but also has better robustness, solves the problems of poor robustness and low invisibility of a color image digital watermarking algorithm, and is suitable for occasions of digital media copyright protection with strong robustness and high invisibility.

The invention discloses a robust image watermarking method based on a fast and high-precision Pseudo-Zernike moment. According to the method, the defects of the existing orthogonal moment watermarking algorithm in precision and calculation efficiency are overcome, the copyright of the digital image is effectively protected, and meanwhile, the resistance of the watermark to common image processing attacks and geometric attacks is improved. The method specifically comprises the following steps: (1) adopting a polar coordinate pixel division and nearest neighbor interpolation algorithm, deducing an analytical solution of a Pseudo-Zernike moment in combination with a Newton-Leibniz formula, and calculating a rapid and high-precision Pseudo-Zernike moment by utilizing symmetric and antisymmetric characteristics of a primary function; (2) screening Pseudo-Zernike moments meeting conditions, carrying out normalization processing, encrypting a watermark sequence by adopting tent chaotic mapping, and embedding the encrypted watermark into the amplitude of the normalization moment through an improved quantitative modulation strategy to obtain a moment value of the embedded watermark; and (3) carrying out inverse Pseudo-Zernike moment transformation on the moment embedded with the watermark, reconstructing to obtain an image with the watermark, and carrying out range correction on a pixel value to generate a final watermark image.

This invention provides a data synchronization method and apparatus. Through a visual interface, the data source and target cluster index parameters are configured. A data synchronization task is created based on the data source and index parameters. The task's time period and triggering rules are set. The task is scheduled according to these rules. An incremental watermark algorithm is used to read data from the data source. Write nodes are dynamically allocated based on load. A search interface combining HTTP / Socket dual protocols and local caching is provided to query data for business scenarios. This improves development efficiency, ensures data consistency and reliability, enhances resource utilization, and reduces data synchronization time. It significantly improves the data synchronization efficiency and system stability of distributed systems and is suitable for high-concurrency, multi-source, heterogeneous data synchronization scenarios.

The invention provides a convolutional neural network model integrity verification method based on an adaptive watermarking algorithm, and the method is characterized in that the method comprises the following steps: carrying out the watermark carrier selection based on a structured pruning method, and obtaining a watermark carrier sequence; performing watermark embedding on the watermark carrier sequence based on an adaptive watermark algorithm and a histogram shift algorithm to obtain a model with a first watermark and an extraction key; according to the extraction key, using a histogram shift algorithm to extract a first watermark in the model, and reconstructing the model; calculating the information abstract of the reconstructed model to obtain a second watermark; and verifying the integrity of the model according to a comparison result of the first watermark and the second watermark. According to the method, the technical problems that when the integrity of the convolutional neural network is verified by using the fragile model watermark, the precision of the convolutional neural network is lost and the watermark capacity is difficult to control due to an existing watermark embedding mode can be solved.

The invention discloses a traceable electronic license secret security sharing method, which comprises the following steps of: S1, writing an electronic license which needs to be safely shared into an exclusive signature, and adding a watermark by adopting an error-correctable robust watermark algorithm; s2, encrypting the electronic certificate with the exclusive signature to generate a secret image S; and S3, dividing the secret image S into n parts of core shadow sub-images and non-core shadow sub-images, and respectively storing the core shadow sub-images and the non-core shadow sub-images with participants with different credibility. The invention also correspondingly provides a traceable electronic license secret security sharing system. According to the method and the device, the problems of privacy leakage of the electronic license in the transmission process of different users and responsibility investigation after leakage are solved.

The application provides a multi-bit audio watermarking method based on phase distribution and high-efficiency bit mapping, and belongs to the field of audio information hiding. The application divides multiple independent phase subintervals based on the natural distribution characteristics of phases, and can realize one-to-many mapping from single feature to multi-bit watermarking by constructing unique phase features in the phase subintervals corresponding to watermarking information. In addition, the application introduces an optimization strategy, and can realize automatic phase redistribution. The application can significantly reduce the number of subspaces / feature modes required for multi-bit embedding, and reduce the design difficulty of the multi-bit watermarking algorithm. In addition, since an automatic phase redistribution algorithm based on optimization constraints is adopted, the algorithm can more effectively balance the inaudibility and robustness.

The invention discloses an image watermarking method based on Hessenberg decomposition and three-dimensional quantization index modulation. According to the method, matrix decomposition is carried out on an image block through Hessenberg decomposition, first three coefficients with absolute values arranged in a descending order are selected from a Hessenberg matrix of the image block to carry out three-dimensional joint quantization index modulation so as to improve the robustness of the watermarking method, the quantization step size is adaptively adjusted by utilizing RGB channel correlation so as to keep the invisibility of a watermark, a four-dimensional Chen's chaos encryption system is introduced, and the robustness of the watermarking method is improved. And the key space is expanded to 2434. The method solves the problems of weak robustness, low invisibility and poor security of a color image digital watermarking algorithm, and is suitable for occasions of digital media copyright protection with strong robustness, high invisibility and high security.

This invention combines the advantages of high speed in spatial domain digital watermarking algorithms and high robustness in frequency domain digital watermarking algorithms, disclosing a fusion-domain blind watermarking method based on graph transform and particle swarm optimization (PSO) algorithm. Based on the data correlation-removing properties of graph transform and the optimal solution-obtaining properties of PSO, this invention first obtains the first frequency domain coefficients of the image patch after graph transform in the spatial domain. Then, these coefficients are used in the spatial domain to complete the embedding and blind extraction of the digital watermark. Finally, PSO is used for optimization to select the optimal quantization step size. This invention can embed a color image digital watermark into a color host image, possessing not only good watermark concealment and strong robustness, but also good real-time performance and security. It solves the problem of slow operation speed in large-capacity color image digital watermarking and is suitable for fast, efficient, and secure digital media copyright protection.

The invention provides an imperceptible watermark attack method based on Transform and prompt guidance, and belongs to the technical field of digital information security. The method comprises the following steps: constructing a training data set through an existing data set and a QPHFMs watermark algorithm, and uniformly adjusting the sizes of a watermark-containing image and an original watermark-free image to obtain standardized training data; the method comprises the following steps: constructing an imperceptible watermark attack network TFP-WAN comprising a Transform encoder-decoder framework and a dynamic prompt module, wherein the dynamic prompt module comprises a PGM and a PIM; mAE is used as a loss function in the TFP-WAN; inputting the standardized training data into the TFP-WAN for iterative training until a preset number of iterations is reached, and obtaining a trained watermark attack model; and inputting a watermark-containing image to be attacked into the watermark attack model, and outputting an attacked image. According to the method, the attack on the robust watermarking algorithm is realized with a relatively high error rate.

The application discloses an audio robust watermark generation method and a terminal based on an embedding strength prediction network, and the method comprises the following steps: acquiring audio data, performing frame division and watermark embedding processing on the audio data, and generating a corresponding three-dimensional coefficient matrix; inputting the three-dimensional coefficient matrix into an embedding strength prediction network, and outputting a corresponding embedding strength factor; adjusting the embedding strength of the watermark according to the embedding strength factor, and outputting an audio frame embedded with watermark information; the application proposes a new self-adaptive audio robust watermark generation method based on an embedding strength prediction network, and improves the imperceptibility and robustness of the watermark algorithm.

The present invention proposes a multi-size image robust watermarking method and system based on an attention mechanism, which relates to the field of digital image watermarking. The method comprises the following steps: preprocessing an input color image and extracting an image feature tensor through an image feature extraction network; performing Arnold scrambling on the watermark information to be embedded by using the watermark feature extraction network to generate a watermark feature tensor of the same size as the image feature tensor; deeply fusing the image feature tensor and the watermark feature tensor through the watermark embedding network to generate a color watermarked image; performing a simulated attack on the color watermarked image by using a simulated attack network to generate an attacked image; and inputting the attacked image into the watermark extraction network to extract the watermark information. The present invention utilizes a multi-head attention mechanism to capture richer feature information, utilizes the attention mechanism to enhance useful channel features, and suppresses unimportant channel features, thereby improving the imperceptibility of the method and the embedding capacity and robustness of the watermarking algorithm.

The invention provides a high-performance AI visual stealth watermarking method and system based on video inter-frame differences, and the method comprises the steps: improving and enhancing a semantic segmentation part network structure in an embedded model of an existing stealth watermarking algorithm, recognizing an inter-frame motion change rate in a video frame image through employing a GRU algorithm, and achieving the purpose that only an area exceeding a threshold value in the image is inferred and calculated again. The AIGC model is utilized to generate an image training set for simulating dynamic change differences, and a joint feedback training mechanism clamped by human visual evaluation is utilized to achieve the effects of greatly reducing performance consumption and computing power requirements and ensuring extraction accuracy and visual concealment at the same time.

The application discloses a strong robust frequency domain symmetrical image watermarking method for copyright confirmation, utilizes the base image features of different frequency domain coefficients, modulates specific frequency domain coefficients to realize the embedding of symmetrical watermarking, enhances the robustness of the watermarking against large-area cutting attacks by constructing symmetrical peaks with consistent number of different resolution images, and designs a watermark preprocessing module meeting the Kerckhoffs principle to guarantee the security of the watermarking algorithm under the open condition. The application can effectively resist various signal processing attacks and geometric attacks, still has high-precision watermark resynchronization ability when suffering various attacks, enhances the robustness of the watermarking against various attacks while guaranteeing the high perceptual sensitivity of the image, is stronger in the robustness against large-area cutting, and is stable in performance and good in generalization when applied to different resolution images.

The invention discloses a robust zero watermarking algorithm based on NSCT and UPEMD-PHT. The method comprises the following steps: step 1, carrying out Tent chaotic encryption on a watermark image; step 2, embedding a watermark; step 3, extracting a watermark; and 4, decrypting the watermark image. According to the method, before a carrier image is embedded, encryption of a watermark image is carried out based on Tent chaotic mapping so as to ensure the security of the watermark, non-subsampled contourlet transform (NSCT) is carried out on a host image, low-frequency information is extracted, and then uniform phase empirical mode decomposition (UPEMD) and polar harmonic transform (PHT) are carried out on the low-frequency information so as to complete watermark embedding. According to the invention, a chaotic mapping algorithm and a novel watermarking technology are organically combined, and the method for embedding the digital zero watermark in the digital image is realized.

Disclosed is a commutative encryption and watermarking method utilizing deoxyribonucleic acid (DNA) base complementary dynamic encryption for a high-definition map. According to the method, on the basis of a DNA base pair complementary principle and data features of the high-definition map in the OpenDRIVE format, a commutative encryption and watermarking algorithm based on base pair complementary dynamic encryption for a high-definition map. In an encryption method, cubic curve parameters configured to denote elevation elements and lane widths in data are encrypted through a nucleotide base complementary principle in DNA. In a watermarking method, a number of elevation elements and lanes is not influenced by encryption, and the invariant feature is used to generate a zero-watermark. Interchangeability of encryption and watermarking is obtained.

The present invention relates to the technical field of copyright protection, and discloses a copyright protection method for self-supervised learning visual models, including the following steps: Step S1: Before watermark generation, it is first necessary to use the encoder E pre-trained through self-supervised learning θ to generate an adversarial perturbation w adv and use this as the watermark; Step S2: Watermark embedding, after generating w adv the next step is to embed w adv into the pre-trained encoder E θ ; Step S3: Watermark verification, when the encoder copyright owner verifies whether a suspicious encoder has infringed the intellectual property rights of its watermarked encoder, its copyright can be verified in two scenarios: the white-box scenario and the black-box scenario. This copyright protection method for self-supervised learning visual models is different from the traditional watermark algorithm for end-to-end classification models. This algorithm does not require prior knowledge of downstream tasks; therefore, this algorithm can verify the ownership of the encoder in the white-box or even more stringent black-box scenario.

The present invention relates to the field of image processing, and particularly to a robust image watermarking algorithm based on dual channels. After performing DWT on the U channel, subbands are selected for blocking, then DCT is carried out. The values in the watermark sequence are embedded by selecting the DCT mid-frequency coefficients in each block. Then, after performing inverse DCT transformation and then inverse DWT transformation, the U channel image with the embedded watermark is obtained; after performing DWT transformation on the V channel, the selected subbands are subjected to DCT transformation. The DCT coefficients are subjected to SVD to obtain the singular value matrix S. The grayscale image of the second watermark image is subjected to SVD decomposition to obtain the singular value matrix S w , and S w is embedded into the S matrix according to the embedding rules to complete the watermark embedding. Then, inverse SVD decomposition and inverse DCT transformation are carried out to obtain a new wavelet subband, and inverse DWT transformation is performed on it to obtain the V channel image with the embedded watermark. Finally, the image with the embedded watermark is synthesized. The present invention uses different algorithms for watermark image embedding and extraction, making the image with the embedded watermark have stronger robustness and also ensuring inappreciability.

The present application relates to the technical field of computer application, specifically to a traceable deep learning model black box watermarking scheme, which proposes a black box watermarking scheme capable of verifying the identity of a deep model. The main method is to generate a trigger set using an image watermarking algorithm to form an efficient, traceable and reproducible deep model black box watermarking scheme, which solves the model identity recognition problem and the current intelligent model watermarking scheme and ownership claim attack problem. The trigger set picture generated by adding random noise or using a deep network is difficult to reproduce. Even in the case of being able to reproduce, it is still impossible to trace and verify the identity. In the case of trigger set loss, the trigger set can still be reproduced through the key and the original trigger picture. This effectively improves the reliability of the scheme.

The invention provides a Gaussian splash model reversible watermark generation method and device, and relates to the technical field of watermark algorithms, and the method comprises the steps: selecting a target seal model; selecting a reference Gaussian primitive from the Gaussian splash model to be protected, and determining a reference position based on the covariance of the reference Gaussian primitive; selecting a plurality of nearest neighbor Gaussian primitives in a neighborhood by taking the reference position as a center to construct a local primitive set; performing feature decoupling on the local primitive set to obtain a plurality of geometric appearance features, and adding nonlinear perturbation to the plurality of geometric appearance features to obtain perturbation geometric appearance features; and performing transformation processing on the target seal model by using the disturbance geometric appearance features, and embedding the transformed target seal model into the to-be-protected Gaussian splash model to obtain a Gaussian splash model carrying the dominant watermark seal. By adopting the reversible watermark generation method and device of the Gaussian splash model, the problems that the precision of the 3DGS model cannot be recovered after the watermark is removed and the watermark generation complexity is high are solved.

The application discloses an unmanned system network attack detection method based on a recursive watermark mechanism, first, an unmanned system model, a controller model and a detector model under network attack are established, second, a network attack model with non-continuous characteristics is established according to the unmanned system model, then, a recursive watermark mechanism based on an event trigger mechanism of a detector alarm is introduced, finally, the detection effect of the detector and the system performance loss are calculated according to the unmanned system model and the recursive watermark parameters proposed in the application. Compared with the traditional periodic watermark method, the detection rate of the recursive watermark algorithm is higher under the same performance loss, which is of great significance for guiding the deployment of unmanned system security defense measures and saving system performance loss, thereby realizing the guarantee of defense effect and the reduction of performance system loss.