A beacon lossless construction method for generative images

Abstract

The application discloses a kind of beacon lossless construction methods for generative image.The beacon lossless construction method for generative image includes the following steps: S1 generative initial image set acquisition;S2 generative image beacon parameter acquisition;S3 generative image target beacon selection;S4 generative image target beacon position selection;S5 generative image target beacon fault tolerance parameter acquisition;S6 generative image target beacon embedding;S7 generative image target beacon verification.The application determines the position of generative image target beacon and embedded image according to generative image beacon parameter and generative image target beacon position parameter, analyzes the embedding after adding fault tolerance parameter in beacon embedding process, calculates generative image beacon loss by extracting beacon to verify generative image target beacon, achieves the effect of standardizing the evaluation of beacon construction method for generative image, solves the problem of non-standard hidden information construction evaluation method in the prior art.

Description

Technical Field

[0001] This invention relates to the field of computer vision technology, and in particular to a lossless beacon construction method for generative images. Background Technology

[0002] Image generation models are broadly classified into unconditional generation and conditional generation. Unconditional generation primarily includes Variational Autoencoders (VAEs), which generate images whose dimensions remain unchanged after dimensionality upscaling and downscaling operations. Conditional generation mainly includes Generative Adversarial Networks (GANs), which generate realistic images by training generator and discriminator networks. With increasing demands for image authentication, security, copyright protection, and privacy protection, image information hiding technologies are developing rapidly.

[0003] Image information hiding techniques are primarily achieved through two methods: embedded beacons and digital watermarking. Beacon construction refers to embedding a beacon carrying specific information at a specific location within media such as images, audio, and video without affecting the quality of the media. This is used for image identification and verification, as well as for anonymization and image authentication. Digital watermarking is a technique that embeds hidden information into images and can be used for image authentication, copyright protection, and tracking.

[0004] For example, the invention patent with publication number CN114777757A discloses a beacon map construction method, apparatus, device, and storage medium, including: acquiring the beacon position measured by the measuring device at the (i-1)th station; acquiring the first pose constraint relationship of the measuring device at the i-th station relative to the (i-1)th station based on the pose change information of the station change; acquiring the second pose constraint relationship of the i-th station relative to the beacon based on the beacon pose collected at the i-th station and combined with the determined located beacon; determining the error equation of the i-th station based on the first pose constraint relationship and the second pose constraint relationship; optimizing the error equation to determine the position of the i-th station; and determining the beacon map based on the determined station position.

[0005] For example, the invention patent with publication number CN103455966A discloses a digital watermark embedding device, a digital watermark embedding method, and a digital watermark detection device. The digital watermark embedding device includes: an interface unit for acquiring video data and digital watermark information; and a processing unit for embedding the digital watermark information into the video data. The processing unit is configured such that the area of ​​the watermark pattern, formed by a plurality of pixels having specified values ​​and superimposed on each image contained in the video data, changes periodically over time according to the value of the symbol contained in the digital watermark information, and uses the specified values ​​to correct the value of each pixel contained in the area where each image in the video data and the corresponding watermark pattern overlap with each other.

[0006] However, in the process of implementing the inventive technical solution in the embodiments of this application, the inventors of this application discovered that the above-mentioned technology has at least the following technical problems:

[0007] In existing technologies, beacon construction uses existing results and error equations to optimize and obtain the next beacon, while digital watermark construction uses interface units and processing units to acquire video data, digital watermarks, and embed watermark information. However, there is a problem with the lack of standardized evaluation methods for hidden information construction. Summary of the Invention

[0008] This application provides a lossless beacon construction method for generative images, which solves the problem of non-standard evaluation methods for hidden information construction in the prior art and realizes a standardized evaluation of the lossless beacon construction method for generative images.

[0009] This application provides a lossless beacon construction method for generative images, comprising the following steps: S1, obtaining a generative initial image set: based on the principle of image generation, selecting an image generation model to obtain generative initial images, and constructing a generative initial image set; S2, obtaining generative image beacon parameters: analyzing all generative initial images in the generative initial image set to obtain image feature parameters of the generative images, and determining generative image beacon parameters based on the image feature parameters of the generative images; S3, selecting generative image target beacons: analyzing the relationship between generative image beacon parameters of all generative initial images in the generative initial image set to obtain generative image target beacon parameters, and substituting the generative image target beacon parameters into the generative image target beacon formula to select generative image target beacons; S4, selecting generative image target beacon positions: analyzing the generative initial image beacon parameters of all generative initial images in the generative initial image set to obtain generative image target beacon position parameters, and substituting the generative image target beacon position parameters into the generative image target beacon formula. The generative image target beacon position index is calculated using the image target beacon position index formula to obtain the generative image target beacon position; S5, Generative image target beacon fault tolerance parameter acquisition: The process of embedding the generative image target beacon into the generative initial image is analyzed to obtain the generative image target beacon fault tolerance parameter; S6, Generative image target beacon embedding: The generative image target beacon is analyzed and the generative image target beacon fault tolerance parameter is added. The generative image target beacon is embedded at the generative image target beacon position in the generative initial image according to the beacon embedding principle to obtain the generative image target beacon image; S7, Generative image target beacon verification: The beacon is extracted from the generative image target beacon image using the beacon extraction method to obtain the generative image reference beacon. The correlation between the generative image reference beacon and the generative image target is analyzed. The correlation between the generative image target beacon image and the generative initial image is analyzed to obtain the beacon loss parameter. The beacon loss parameter is substituted into the generative image beacon loss formula to calculate the generative image beacon loss and verify the degree of loss of the generative image beacon embedding.

[0010] Furthermore, the generative initial image set in S1 is a collection of generative initial images, and the generative initial images are high-quality images.

[0011] Furthermore, the generative image beacon parameters in S2 include a grayscale set, a color channel set, an image resolution set, an image texture, and an image frequency, specifically as follows: The color channel set is a beacon combination of the red, green, and blue channel values ​​at a selected region. The specific steps for obtaining this set are: based on the red, green, and blue channel values, the color channel set is obtained using a formula, specifically the following formula: Where Color (color channel), Color (red channel), Color (green channel), and Color (blue channel) are the values ​​of the color channel, red channel, green channel, and blue channel, respectively. The integer part is used for rounding. α1, β1, and γ1 are the coefficients of the red, green, and blue channel values, respectively, and their calculation formulas are as follows: and e is a natural constant; the image resolution set is the numerical value of the image resolution.

[0012] Furthermore, the generative image target beacon parameters in S3 are a color channel set, an image resolution set, and a grayscale frequency set, specifically as follows: the grayscale frequency set is the set of frequencies of each gray level in the generative initial image, obtained from the grayscale histogram; the specific steps for obtaining the generative image target beacon are as follows: based on the color channel set, image resolution set, and grayscale frequency set, the generative image target beacon is obtained through a formula, specifically the formula XB(target beacon) = {Color(color channel)}∪{Resolution(image resolution set)}∩{Gray(grayscale frequency set)}, where XB(target beacon), Color(color channel), Resolution(image resolution set), and Gray(grayscale frequency set) are the generative image target beacon, the color channel set of the generative initial image, the image resolution set of the generative initial image, and the grayscale frequency set of the generative initial image, respectively.

[0013] Furthermore, in S4, the generative image target beacon position parameters are image texture and image frequency. The generative image target beacon position index is calculated from the generative image target beacon position parameters, specifically as follows: The image texture is measured using entropy and angular second moment. The specific steps are as follows: Based on the entropy value E(x,y) and the angular second moment ASM(x,y), the image texture W is obtained through the formula. 纹理 (x,y), the specific calculation formula is as follows e is the natural constant;

[0014] The specific steps for obtaining the generative image target beacon location index are as follows: based on the image texture W... 纹理 (x, y) is used to obtain the target beacon location index of the generative image through a formula, the specific calculation formula is as follows: Where L(x,y) and P 频率 (x, y) represent the generative image target beacon position index and image frequency at (x, y) in the image, respectively; the generative image target beacon position is the region with the largest generative image target beacon position index in the reference region of the generative initial image, and the generative image reference region is the top left vertex, bottom left vertex, top right vertex, bottom right vertex and centroid of the generative initial image.

[0015] Furthermore, the generative image target beacon fault tolerance parameters in S5 are the generative image transformation fault tolerance parameters and the generative image target beacon encoding check code, specifically as follows: the generative image transformation fault tolerance parameters are the target frequency threshold of the generative initial image; the generative image target beacon encoding check code is the forward error correction code and the cyclic redundancy code.

[0016] Further, the generative target beacon image is obtained in S6, specifically through the following process: S61, frequency domain image acquisition: the generative initial image is converted from the spatial domain to the frequency domain through Fourier transform, and the generative image transformation tolerance parameter is set to obtain the frequency domain image; S62, target beacon frequency domain data acquisition: Huffman coding is performed on the target beacon in the generative image, and a generative image target beacon encoding check code is added to obtain the target beacon frequency domain data in numerical form; S63, target beacon frequency domain data embedding: the target beacon frequency domain data is embedded into the target beacon position of the generative image in the frequency domain image to obtain the target beacon frequency domain image; S64, generative target beacon image acquisition: the target beacon frequency domain image is converted from the frequency domain to the spatial domain through inverse Fourier transform to obtain the generative target beacon image.

[0017] Furthermore, the comparison and verification with the generative image target beacon in S7 is carried out as follows: S71, Generative image reference beacon extraction: The generative target beacon image is converted from the spatial domain to the frequency domain through Fourier transform, the frequency domain data of the target beacon is extracted, and then the frequency domain data of the target beacon is decoded through Huffman decoding to obtain the generative image reference beacon; S72, Calculate image loss: The relationship between the generative target beacon image and the generative initial image is analyzed to obtain image loss parameters, and the image loss of the generative target beacon image and the generative initial image is calculated using the image loss formula; S73, Calculate beacon loss: The relationship between the generative image reference beacon and the generative image target beacon is analyzed to obtain beacon loss parameters, and the beacon loss of the generative image reference beacon and the generative image target beacon is calculated using the beacon loss formula; S74, Calculate generative image beacon loss: The generative image beacon loss is obtained from the image loss and the beacon loss.

[0018] Furthermore, the image loss and beacon loss are calculated as follows: the image loss parameters are typical evaluation indicators of the similarity between two images, namely structural similarity, perceptual hashing, and peak signal-to-noise ratio; the specific steps for obtaining the image loss are: based on the image loss parameters, the image loss is obtained through a formula, the specific calculation formula being as follows: Loss 图像 SSIM 图像 pHash 图像 and PNSR 图像The parameters are: image loss, structural similarity, perceptual hashing, and peak signal-to-noise ratio (PSNR) for the generative target beacon image and the generative initial image, respectively, where e is the natural constant. The beacon loss parameters are typical evaluation indicators of the correlation between the two sets of data, namely linear correlation, variance, standard deviation, and kurtosis. The specific steps for obtaining the beacon loss are: based on linear correlation, variance, standard deviation, and kurtosis, the beacon loss is obtained using the following formula: Loss 信标 R1 信标 R2 信标 R3 信标 and R4 信标 These are the beacon loss, beacon linear correlation, beacon variance correlation, beacon standard deviation correlation, and beacon kurtosis correlation of the generative image reference beacon and the generative image target beacon, respectively. The specific steps for obtaining the beacon variance correlation between the generative image reference beacon and the generative image target beacon are as follows: based on the variances of the generative image reference beacon and the generative image target beacon, the beacon variance correlation R² between the generative image reference beacon and the generative image target beacon is obtained using the formula. 信标 The specific calculation formula is as follows: Sd 目标信标 and Sd 参考信标 These are the variances of the generative image target beacon and the generative image reference beacon, respectively. The specific steps for obtaining the beacon standard deviation correlation between the generative image reference beacon and the generative image target beacon are as follows: based on the standard deviations of the image reference beacon and the generative image target beacon, the beacon standard deviation correlation R3 between the generative image reference beacon and the generative image target beacon is obtained using the formula. 信标 The specific calculation formula is R3 信标 The specific calculation formula is as follows: Cov 目标信标 and Cov 参考信标 These are the standard deviations of the generative image target beacon and the generative image reference beacon, respectively. The specific steps for obtaining the beacon kurtosis correlation between the generative image reference beacon and the generative image target beacon are as follows: based on the kurtosis of the generative image reference beacon and the generative image target beacon, the beacon kurtosis correlation R4 between the generative image reference beacon and the generative image target beacon is obtained using the formula. 信标 The specific calculation formula is as follows: Where K 目标信标 and K 参考信标 These are the kurtosis of the target beacon in the generative image and the kurtosis of the reference beacon in the generative image, respectively.

[0019] Furthermore, the specific steps for obtaining the generative image beacon loss are as follows: based on the image loss and variance loss, the generative image beacon loss is obtained through a formula, specifically the following formula: Loss 总体 This is the generative image beacon loss.

[0020] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0021] 1. By acquiring a generative initial image set, obtaining generative image beacon parameters, selecting generative image target beacons, selecting generative image target beacon positions, and acquiring generative image target beacon fault tolerance parameters, generative image target beacons are embedded at the generative image target beacon positions in the generative initial image to obtain generative image target beacon images. The loss degree of generative image beacon embedding is verified, thereby realizing a standardized evaluation of the lossless beacon construction method for generative images, effectively solving the problem of non-standard evaluation methods for hidden information construction in existing technologies.

[0022] 2. By using Fourier transform to convert the initial generative image from the spatial domain to the frequency domain to obtain the frequency domain image, and by Huffman coding of the target beacon in the generative image to obtain the numerical form of the target beacon frequency domain data, the target beacon frequency domain data is embedded into the target beacon position of the generative image in the frequency domain image to obtain the target beacon frequency domain image. This achieves accurate and fast conversion of the target beacon frequency domain image from the frequency domain to the spatial domain.

[0023] 3. Generative image reference beacons are extracted from generative target beacon images through Fourier transform and Huffman decoding. Then, the image loss and beacon loss in the beacon embedding of generative images are calculated using image loss formula and beacon loss formula to obtain the generative image beacon loss, thereby realizing a numerical measurement of the degree of beacon embedding loss of generative images. Attached Figure Description

[0024] Figure 1 A flowchart illustrating a lossless beacon construction method for generative images provided in this application embodiment;

[0025] Figure 2 This is a flowchart of a generative target beacon image acquisition process provided in an embodiment of this application;

[0026] Figure 3 This is a flowchart illustrating the generative image target beacon verification process provided in an embodiment of this application. Detailed Implementation

[0027] This application provides a lossless beacon construction method for generative images, which solves the problem of non-standard evaluation methods for hidden information construction in the prior art. By determining the position of the target beacon and the embedded image in the generative image through beacon parameters and target beacon position parameters in the generative image, analyzing the beacon embedding process, adding fault tolerance parameters before embedding, and extracting the beacon to calculate the beacon loss in the generative image to verify the target beacon in the generative image, a standardized evaluation of the lossless beacon construction method for generative images is achieved.

[0028] The technical solution in this application embodiment aims to address the problem of non-standard methods for constructing and evaluating hidden information in the existing technology. The overall approach is as follows:

[0029] By determining the positions of the generative image target beacon and the embedded image using generative image beacon parameters and generative image target beacon position parameters, analyzing the beacon embedding process after adding fault tolerance parameters, extracting the beacon, and calculating the generative image beacon loss, the generative image target beacon is verified, achieving the effect of standardized evaluation of the lossless beacon construction method for generative images.

[0030] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0031] like Figure 1The diagram shows a flowchart of a lossless beacon construction method for generative images provided in this application embodiment. The method includes the following steps: S1, obtaining a generative initial image set: Based on the principle of image generation, an image generation model is selected to obtain generative initial images, thus constructing a generative initial image set; S2, obtaining generative image beacon parameters: Analyzing all generative initial images in the generative initial image set to obtain image feature parameters of the generative images, and determining generative image beacon parameters based on the image feature parameters of the generative images; S3, selecting generative image target beacons: Analyzing the relationship between generative image beacon parameters of all generative initial images in the generative initial image set to obtain generative image target beacon parameters, and substituting the generative image target beacon parameters into the generative image target beacon formula to select generative image target beacons; S4, selecting generative image target beacon positions: Analyzing the generative initial image beacon parameters of all generative initial images in the generative initial image set to obtain generative image target beacon position parameters, and selecting the generative image target beacon positions... S5. Obtaining the Generative Image Target Beacon Position Index: Analyze the process of embedding the generative image target beacon into the generative initial image to obtain the generative image target beacon position; S6. Generative Image Target Beacon Embedding: Analyze the generative image target beacon and add the generative image target beacon fault tolerance parameters. Embed the generative image target beacon at the generative image target beacon position in the generative initial image according to the beacon embedding principle to obtain the generative image target beacon image; S7. Generative Image Target Beacon Verification: Use the beacon extraction method to extract the beacon from the generative image target beacon image to obtain the generative image reference beacon. Analyze the correlation between the generative image reference beacon and the generative image target, analyze the correlation between the generative image target beacon image and the generative initial image to obtain the beacon loss parameters. Substitute the beacon loss parameters into the generative image beacon loss formula to calculate the generative image beacon loss and verify the degree of loss of the generative image beacon embedding.

[0032] In this embodiment, the evaluation of the lossless beacon construction method for generative images is standardized by implementing the following steps: obtaining generative initial image set, obtaining generative image beacon parameters, selecting generative image target beacons, selecting generative image target beacon positions, obtaining generative image target beacon fault tolerance parameters, embedding generative image target beacons, and verifying generative image target beacons.

[0033] Furthermore, the generative initial image set in S1 is a collection of generative initial images, and the generative initial images are high-quality images.

[0034] In this embodiment, by selecting an image generation model to obtain high-quality generative initial images and constructing a generative initial image set, the universality of the lossless beacon construction method for generative images is enhanced.

[0035] Furthermore, the generative image beacon parameters in S2 are grayscale set, color channel set, image resolution set, image texture, and image frequency, specifically as follows: The color channel set is a beacon combination of the red, green, and blue channel values ​​for a selected region. The specific steps to obtain this set are: based on the red, green, and blue channel values, the color channel set is obtained using a formula, specifically the following formula: Where Color (color channel), Color (red channel), Color (green channel), and Color (blue channel) are the values ​​of the color channel, red channel, green channel, and blue channel, respectively. The integer part is used for rounding. α1, β1, and γ1 are the coefficients of the red, green, and blue channel values, respectively, and their calculation formulas are as follows: and e is a natural constant; the image resolution set is the numerical value of the image resolution.

[0036] In this embodiment, resolution is one of the important indicators for measuring image quality. High-resolution images have more details and clarity. By analyzing all generative initial images in the generative initial image set, the image feature parameters of the generative image are obtained, and then the beacon parameters of the generative image are determined, thereby enhancing the uniqueness of the lossless beacon construction method for generative images.

[0037] Furthermore, the generative image target beacon parameters in S3 are the color channel set, image resolution set, and grayscale frequency set, as follows: The grayscale frequency set is the set of frequencies of each gray level in the generative initial image, obtained from the grayscale histogram, where grayscale levels are on the horizontal axis and the frequency or number of occurrences is on the vertical axis; The specific steps for obtaining the generative image target beacon are as follows: Based on the color channel set, image resolution set, and grayscale frequency set, the generative image target beacon is obtained through the formula, specifically the formula XB(target beacon) = {Color(color channel)}∪{Resolution(image resolution set)}∩{Gray(grayscale frequency set)}, where XB(target beacon), Color(color channel), Resolution(image resolution set), and Gray(grayscale frequency set) are the generative image target beacon, the color channel set of the generative initial image, the image resolution set of the generative initial image, and the grayscale frequency set of the generative initial image, respectively.

[0038] In this embodiment, by analyzing the relationship between generative image beacon parameters of all generative initial images in the generative initial image set, the generative image target beacon parameters are obtained, and then the generative image target beacon is selected, thereby enhancing the beacon uniqueness for generative images.

[0039] Furthermore, in S4, the generative image target beacon position parameters are image texture and image frequency. The generative image target beacon position index is calculated from the generative image target beacon position parameters, as follows: Image texture is measured using entropy and angular second moment. The specific steps are as follows: Based on the entropy value E(x,y) and angular second moment ASM(x,y), the image texture W is obtained through the formula. 纹理 (x,y), the specific calculation formula is as follows e is a natural constant; the specific steps for obtaining the generative image target beacon position index are as follows: based on the image texture W... 纹理 (x, y) is used to obtain the target beacon location index of the generative image through a formula, the specific calculation formula is as follows: Where L(x,y) and P 频率 (x,y) represent the generative image target beacon position index and image frequency at (x,y) in the image, respectively; the generative image target beacon position is the region with the largest generative image target beacon position index in the reference region of the generative initial image, and the generative image reference region is the top left vertex, bottom left vertex, top right vertex, bottom right vertex and centroid of the generative initial image.

[0040] In this embodiment, the generative image target beacon position is obtained by analyzing the generative initial image beacon parameters and the generative image target beacon position index of all generative initial images in the generative initial image set. This improves the normality of the selection of the generative image target beacon embedding position and reduces the loss.

[0041] Furthermore, the generative image target beacon fault tolerance parameters in S5 are the generative image transformation fault tolerance parameters and the generative image target beacon encoding check code, as follows: The generative image transformation fault tolerance parameters are the target frequency threshold of the generative initial image, which can filter out noise and false detections; the generative image target beacon encoding check code are the forward error correction code and the cyclic redundancy code, which can reduce errors that may be introduced during transmission or storage.

[0042] In this embodiment, the error tolerance parameters of the generative image target beacon are obtained by analyzing the process of embedding the generative image target beacon into the generative initial image, thereby reducing the error rate and improving the robustness of the generative image target beacon embedding process.

[0043] Furthermore, such as Figure 2The diagram shows a flowchart of the generative target beacon image acquisition process provided in this application embodiment. The generative target beacon image is obtained in step S6, and the specific process is as follows: S61, Frequency domain image acquisition: The initial generative image is converted from the spatial domain to the frequency domain using Fourier transform. A fault-tolerant parameter for the generative image transformation is set to obtain the frequency domain image, which has higher stability. S62, Target beacon frequency domain data acquisition: Huffman coding is performed on the target beacon in the generative image, and a generative image target beacon encoding check code is added to obtain the numerical form of the target beacon frequency domain data. S63, Target beacon frequency domain data embedding: The target beacon frequency domain data is embedded into the target beacon position of the generative image in the frequency domain image to obtain the target beacon frequency domain image. S64, Generative target beacon image acquisition: The target beacon frequency domain image is converted from the frequency domain to the spatial domain using inverse Fourier transform to obtain the generative target beacon image.

[0044] In this embodiment, by analyzing generative image target beacons, adding generative image target beacon fault tolerance parameters, and embedding generative image target beacons, a generative target beacon image is obtained, thereby improving the reliability, stability, and uniqueness of generative image target beacon embedding.

[0045] Furthermore, such as Figure 3 The diagram shows a flowchart of the generative image target beacon verification process provided in this application embodiment. In step S7, the verification is compared with the generative image target beacon. The specific process is as follows: S71, Generative image reference beacon extraction: The generative target beacon image is converted from the spatial domain to the frequency domain using Fourier transform, and the target beacon frequency domain data is extracted. Then, the target beacon frequency domain data is decoded using Huffman decoding to obtain the generative image reference beacon; S72, Image loss calculation: The relationship between the generative target beacon image and the generative initial image is analyzed to obtain the image loss parameters, which are then calculated using the image loss formula. Calculate the image loss of the generated target beacon image and the generated initial image. The larger the image loss value, the easier it is for the generated image beacon to be detected. S73, Calculate the beacon loss: Analyze the relationship between the generated image reference beacon and the generated image target beacon to obtain the beacon loss parameters. Calculate the beacon loss of the generated image reference beacon and the generated image target beacon using the beacon loss formula. The larger the beacon loss value, the worse the stability of the generated image target beacon. S74, Calculate the generated image beacon loss: Obtain the generated image beacon loss from the image loss and the beacon loss.

[0046] In this embodiment, beacons are extracted from the generative target beacon image using a beacon extraction method, the correlation between the generative image reference beacon and the generative image target is analyzed, and the correlation between the generative target beacon image and the generative initial image is analyzed to calculate the generative image beacon loss, thereby verifying the degree of loss of generative image beacon embedding.

[0047] Furthermore, the image loss and beacon loss are calculated as follows: The image loss parameters are typical evaluation metrics for the similarity between two images, namely structural similarity, perceptual hashing, and peak signal-to-noise ratio; the specific steps for obtaining the image loss are: based on the image loss parameters, the image loss is obtained through a formula, the specific calculation formula is as follows: Loss 图像 SSIM 图像 pHash 图像 and PNSR 图像 The parameters are: image loss, structural similarity, perceptual hashing, and peak signal-to-noise ratio (PSNR) for the generative target beacon image and the generative initial image, respectively, where e is the natural constant. The beacon loss parameters are typical evaluation metrics for the correlation between the two sets of data, namely linear correlation, variance, standard deviation, and kurtosis. The specific steps for obtaining the beacon loss are: based on linear correlation, variance, standard deviation, and kurtosis, the beacon loss is obtained using the following formula: Loss 信标 R1 信标 R2 信标 R3 信标 and R4 信标 These are the beacon loss, beacon linear correlation, beacon variance correlation, beacon standard deviation correlation, and beacon kurtosis correlation for the generative image reference beacon and the generative image target beacon, respectively. The specific steps for obtaining the beacon variance correlation between the generative image reference beacon and the generative image target beacon are as follows: based on the variances of the generative image reference beacon and the generative image target beacon, the beacon variance correlation R² between the generative image reference beacon and the generative image target beacon is obtained using the formula. 信标 The specific calculation formula is as follows: Sd 目标信标 and Sd 参考信标 These are the variances of the generative image target beacon and the generative image reference beacon, respectively. The specific steps for obtaining the beacon standard deviation correlation between the generative image reference beacon and the generative image target beacon are as follows: Based on the standard deviations of the image reference beacon and the generative image target beacon, the beacon standard deviation correlation R3 between the generative image reference beacon and the generative image target beacon is obtained using the formula. 信标 The specific calculation formula is as follows: Cov 目标信标 and Cov 参考信标 These are the standard deviations of the generative image target beacon and the generative image reference beacon, respectively. The specific steps for obtaining the beacon kurtosis correlation between the generative image reference beacon and the generative image target beacon are as follows: Based on the kurtosis of the generative image reference beacon and the generative image target beacon, the beacon kurtosis correlation R4 between the generative image reference beacon and the generative image target beacon is obtained using the formula. 信标 The specific calculation formula is as follows: Where K 目标信标 and K 参考信标 These are the kurtosis of the target beacon in the generative image and the kurtosis of the reference beacon in the generative image, respectively.

[0048] In this embodiment, image loss and beacon loss are obtained through six indicators: structural similarity, perceptual hashing, peak signal-to-noise ratio, linear correlation, variance, standard deviation, and kurtosis, thereby reducing the lossiness of the lossless beacon construction method for generative images.

[0049] Furthermore, the specific steps for obtaining the generative image beacon loss are as follows: Based on the image loss and variance loss, the generative image beacon loss is obtained using a formula, the specific calculation formula of which is as follows: Loss 总体 This is the generative image beacon loss.

[0050] In this embodiment, the beacon loss of the generative image is calculated by using image loss and beacon loss, thus realizing a numerical measure of the degree of beacon embedding loss of the generative image.

[0051] The technical solutions in the above embodiments of this application have at least the following technical effects or advantages: Compared with the beacon map construction method, apparatus, device and storage medium disclosed in CN114777757A, the embodiments of this application obtain a generative initial image set, generative image beacon parameters, generative image target beacon selection, generative image target beacon position selection and generative image target beacon fault tolerance parameters, thereby embedding the generative image target beacon into the generative initial image to obtain a generative target beacon image, and then verifying the degree of loss of the generative image target beacon embedding; Compared with the digital watermark embedding device, digital watermark embedding method and digital watermark detection device disclosed in CN103455966A, the embodiments of this application obtain a target beacon frequency domain image by acquiring frequency domain image and target beacon frequency domain data, thereby embedding the target beacon frequency domain data into the generative image target beacon position of the frequency domain image to obtain a target beacon frequency domain image, and then converting the target beacon frequency domain image from the frequency domain to the spatial domain through inverse Fourier transform to obtain a generative target beacon image.

[0052] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0053] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0054] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0055] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0056] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0057] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A lossless beacon construction method for generative images, characterized in that, Includes the following steps: S1, Obtaining the Generative Initial Image Set: Based on the principles of image generation, select an image generation model to obtain generative initial images and construct a generative initial image set; S2, Generative Image Beacon Parameter Acquisition: Analyze all generative initial images in the generative initial image set to obtain the image feature parameters of the generative images, and determine the generative image beacon parameters based on the image feature parameters of the generative images; S3, Generative Image Target Beacon Selection: Analyze the relationship between generative image beacon parameters of all generative initial images in the generative initial image set to obtain generative image target beacon parameters, and substitute these parameters into the generative image target beacon formula to select generative image target beacons. The generative image target beacon parameters in S3 are a color channel set, an image resolution set, and a grayscale frequency set, as detailed below: The gray-level frequency set is the set of frequencies of each gray level in the initial generative image, obtained from the gray-level histogram; The specific steps for obtaining the generative image target beacon are as follows: Based on the color channel set, image resolution set, and grayscale frequency set, the generative image target beacon is obtained using a formula, the specific calculation formula being: ,in , , and These are, respectively, the target beacon of the generative image, the color channel set of the generative initial image, the image resolution set of the generative initial image, and the grayscale frequency set of the generative initial image; S4, Generative Image Target Beacon Location Selection: Analyze the generative initial image beacon parameters of all generative initial images in the generative initial image set to obtain the generative image target beacon location parameters. Substitute the generative image target beacon location parameters into the generative image target beacon location index formula to calculate the generative image target beacon location index and obtain the generative image target beacon location. S5, Obtaining the fault tolerance parameters of the generative image target beacon: Analyze the process of embedding the generative image target beacon into the generative initial image, and obtain the fault tolerance parameters of the generative image target beacon; S6, Generative Image Target Beacon Embedding: Analyze the generative image target beacon and add generative image target beacon fault tolerance parameters. Embed the generative image target beacon at the position of the generative image target beacon in the generative initial image according to the beacon embedding principle to obtain the generative target beacon image; S7, Generative Image Target Beacon Validation: Beacons are extracted from the generative target beacon image using a beacon extraction method to obtain a generative image reference beacon. The correlation between the generative image reference beacon and the generative image target is analyzed, as well as the correlation between the generative target beacon image and the generative initial image. The beacon loss parameters are obtained, and the beacon loss parameters are substituted into the generative image beacon loss formula to calculate the generative image beacon loss, thus verifying the degree of loss in the generative image beacon embedding.

2. The method for lossless beacon construction for generative images as described in claim 1, characterized in that: The generative initial image set in S1 is a collection of generative initial images, and the generative initial images are high-quality images.

3. The method for lossless beacon construction for generative images as described in claim 1, characterized in that, The generative image beacon parameters in S2 are grayscale set, color channel set, image resolution set, image texture, and image frequency, as detailed below: The color channel set is a beacon combination of the red, green, and blue channel values ​​for a selected area. The specific steps to obtain it are as follows: Based on the red, green, and blue channel values, the color channel set is obtained using a formula. The specific calculation formula is as follows: ,in , , and These are the color channel, red channel, green channel, and blue channel values, respectively. The integer symbol, , and These are the coefficients for the red, green, and blue channel values, respectively, and their calculation formulas are as follows: , and e is the natural constant; The image resolution set refers to the numerical values ​​of the image resolution.

4. The method for lossless beacon construction for generative images as described in claim 1, characterized in that, In S4, the generative image target beacon position parameters are image texture and image frequency. The generative image target beacon position index is calculated from the generative image target beacon position parameters, as follows: The image texture is measured using both entropy and the second moment of the angle. The specific steps to obtain this texture are as follows: based on the entropy value... Sum of second moments Image texture is obtained through formula. The specific calculation formula is as follows: e is the natural constant; The specific steps for obtaining the generative image target beacon location index are as follows: based on image texture... The generative image target beacon location index is obtained through a formula, the specific calculation formula is as follows: ,in and In the image Generative image target beacon location index and image frequency at the location; The target beacon position in the generative image is the region with the largest generative image target beacon position index in the reference region of the generative initial image. The reference region of the generative image is the top left vertex, bottom left vertex, top right vertex, bottom right vertex, and centroid of the generative initial image.

5. The method for lossless beacon construction for generative images as described in claim 1, characterized in that, The generative image target beacon fault tolerance parameters in S5 consist of generative image transform fault tolerance parameters and generative image target beacon encoding check codes, as detailed below: The generative image transformation fault tolerance parameter is the target frequency threshold of the generative initial image; The generative image target beacon coding check code is a forward error correction code and a cyclic redundancy code.

6. The method for lossless beacon construction for generative images as described in claim 1, characterized in that, The generative target beacon image is obtained in step S6, and the specific process is as follows: S61, Frequency domain image acquisition: The generative initial image is converted from the spatial domain to the frequency domain through Fourier transform, and the generative image transformation tolerance parameters are set to obtain the frequency domain image; S62, Target beacon frequency domain data acquisition: Perform Huffman coding on the generative image target beacon, add generative image target beacon coding check code, and obtain target beacon frequency domain data in numerical form; S63, Target beacon frequency domain data embedding: The target beacon frequency domain data is embedded into the target beacon position of the generative image in the frequency domain image to obtain the target beacon frequency domain image; S64, Generative target beacon image acquisition: The target beacon frequency domain image is converted from the frequency domain to the spatial domain by inverse Fourier transform to obtain a generative target beacon image.

7. The method for lossless beacon construction for generative images as described in claim 1, characterized in that, The comparison and verification with the generative image target beacon in S7 is performed as follows: S71, Generative Image Reference Beacon Extraction: The generative target beacon image is converted from the spatial domain to the frequency domain through Fourier transform, the frequency domain data of the target beacon is extracted, and then the frequency domain data of the target beacon is decoded through Huffman decoding to obtain the generative image reference beacon; S72, Calculate image loss: Analyze the relationship between the generative target beacon image and the generative initial image to obtain image loss parameters, and calculate the image loss between the generative target beacon image and the generative initial image using the image loss formula; S73, Calculate the beacon loss: Analyze the relationship between the generative image reference beacon and the generative image target beacon to obtain the beacon loss parameters, and calculate the beacon loss of the generative image reference beacon and the generative image target beacon using the beacon loss formula; S74, Calculate the generative image beacon loss: The generative image beacon loss is obtained from the image loss and the beacon loss.

8. The method for lossless beacon construction for generative images as described in claim 7, characterized in that, The image loss and beacon loss are calculated in detail as follows: The image loss parameters are typical evaluation indicators of the similarity between two images, namely structural similarity, perceptual hashing, and peak signal-to-noise ratio; The specific steps for obtaining the image loss are as follows: Based on the image loss parameters, the image loss is obtained using a formula, specifically the following formula: ,in , , and denoted as image loss, structural similarity, perceptual hash, and peak signal-to-noise ratio for the generative target beacon image and the generative initial image, respectively, where e is a natural constant; The beacon loss parameters are typical evaluation indicators of the correlation between two sets of data, namely linear correlation, variance, standard deviation and kurtosis; The specific steps for obtaining the beacon loss are as follows: Based on linear correlation, variance, standard deviation, and kurtosis, the beacon loss is obtained using a formula, specifically the following formula: ,in , , , and The values ​​are beacon loss, beacon linear correlation, beacon variance correlation, beacon standard deviation correlation, and beacon kurtosis correlation for generative image reference beacons and generative image target beacons, respectively. The specific steps for obtaining the beacon variance correlation between the generative image reference beacon and the generative image target beacon are as follows: Based on the variances of the generative image reference beacon and the generative image target beacon, the beacon variance correlation between the generative image reference beacon and the generative image target beacon is obtained using a formula. The specific calculation formula is as follows: ,in and These are the variances of the target beacon and the reference beacon in the generative image, respectively. The specific steps for obtaining the beacon standard deviation correlation between the generative image reference beacon and the generative image target beacon are as follows: Based on the standard deviations of the image reference beacon and the generative image target beacon, the beacon standard deviation correlation between the generative image reference beacon and the generative image target beacon is obtained through a formula. The specific calculation formula is as follows: ,in and These are the standard deviations of the target beacon in the generated image and the reference beacon in the generated image, respectively. The specific steps for obtaining the beacon kurtosis correlation between the generative image reference beacon and the generative image target beacon are as follows: Based on the kurtosis of the generative image reference beacon and the generative image target beacon, the beacon kurtosis correlation between the generative image reference beacon and the generative image target beacon is obtained through a formula. The specific calculation formula is as follows: ,in and These are the kurtosis of the target beacon in the generative image and the kurtosis of the reference beacon in the generative image, respectively.

9. A method for lossless beacon construction for generative images as described in claim 8, characterized in that, The specific steps for obtaining the generative image beacon loss are as follows: Based on the image loss and variance loss, the generative image beacon loss is obtained using a formula, the specific calculation formula being: ,in This is the generative image beacon loss.

Images