CN107563155B - Security steganography method and device based on generation of countermeasure network - Google Patents
Security steganography method and device based on generation of countermeasure network Download PDFInfo
- Publication number
- CN107563155B CN107563155B CN201710670786.1A CN201710670786A CN107563155B CN 107563155 B CN107563155 B CN 107563155B CN 201710670786 A CN201710670786 A CN 201710670786A CN 107563155 B CN107563155 B CN 107563155B
- Authority
- CN
- China
- Prior art keywords
- network
- image
- generated
- steganography
- carrier image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Editing Of Facsimile Originals (AREA)
- Image Processing (AREA)
Abstract
The invention relates to a security steganography method and a security steganography device based on a generation countermeasure network. The method comprises the following steps: under the framework of generating a countermeasure network, generating a carrier image in which information is to be embedded through a generation network, and judging the authenticity of the generated carrier image through a judgment network; the carrier image generated by the network is close to a real image through the dynamic game process of generating the network and judging the network; embedding information into a carrier image generated by a generated network; and then, carrying out secondary classification on the input carrier image and the steganographically-classified image by utilizing a steganographic analysis network to obtain the accuracy rate of classifying the carrier image and the steganographically-classified image into an original image and a steganographically-classified image. The carrier image generated by the invention is closer to a real image in vision, the generation speed is higher, and the security of steganography can be improved.
Description
Technical Field
The invention belongs to the technical field of information, relates to an information steganography technology, and particularly relates to a security steganography method and device based on a generation countermeasure network.
Background
The information steganography technology is one of the main branches of information hiding. The information hiding is to hide a group of secret information (an authorized serial number, information or copyright information and the like) in carrier information by using the sensory redundancy of human sense organs to digital signals, so that a possible attacker is difficult to judge whether the secret information exists or not under the condition of not influencing the sensory effect and the use value of host signals, and is more difficult to intercept, thereby ensuring the safety of information transmission. With the development of scientific technology, the information hiding technology becomes a new research hotspot, especially the wide application of digital media technology, so that the information hiding has further development and wider connotation. The invention also discloses a method for hiding information, which is characterized in that digital watermarks, digital signatures and the like for copyright protection are also included in the category of information hiding, and the invention focuses on the application of information hiding in the traditional sense, namely Steganography (Steganography) on digital media, and mainly uses Steganography with images as carriers.
Currently, when one designs a steganographic algorithm, one typically heuristically considers from a steganographic analysis perspective. For example, the secret message should be embedded in more secure image noise and texture areas. In the Denis and Burnaev (ICLR 2016Open review, 2016) work, an original image is generated using DCGAN (Deep Convolutional adaptive Adversarial Networks), and the steganography effect is detected using a separate steganalysis network, the work is performed in the framework of Adversarial learning, the image carrier suitable for steganography is generated by the generation network using the Adversarial property of the DCGAN's discrimination network and the generation network, and whether the generated image carrier can be steganography is evaluated by using the separate steganalysis network.
The method proposed by Denis and Burnaev (ICLR 2016Open review, 2016) has some limitations, and experiments show that the steganographic method is not safe enough. The network in the article is suitable for embedding the same secret key, and when a random secret key is used, the network discrimination effect is poor. In addition, the effect of the steganalysis network is not obvious.
Disclosure of Invention
The invention aims to provide a secure steganography method and a secure steganography device based on a generation countermeasure network, wherein a generated carrier image is closer to a real image in vision, the generation speed is higher, and the security of steganography can be improved.
The method comprises the steps of firstly generating a carrier image, embedding secret information into the carrier image in an off-line mode, inputting the image with the embedded information and the generated carrier image into a steganalysis Network (Gaussian-Neuron Convolutional Neural Network) Network, and classifying and distinguishing the original carrier image and the steganalysis image by the steganalysis Network. Then, information is embedded into the generated carrier image by using a HUGO (high undetected steGO) algorithm, and then the steganographic analysis network is used for distinguishing. Thus, the method of the invention is proved to be effective not only for the embedding of the fixed key, but also for the embedding of the random key.
The technical scheme adopted by the invention is as follows:
a security steganography method based on a generation countermeasure network comprises the following steps:
1) Under the framework of generating a countermeasure network, generating a carrier image in which information is to be embedded through a generation network, and judging the authenticity of the generated carrier image through a judgment network;
2) Through the dynamic game processes of generating a network and judging the network, the carrier image generated by the generated network is close to a real image;
3) The embedding of information into the network-generated carrier image takes place.
Further, the generating countermeasure network is a WGAN.
Further, the dynamic gaming process for generating the network and discriminating the network includes:
a) For the generation of the network, the data it produces are made to coincide as far as possible with the real data, so as to generate a carrier image in which the information is to be embedded;
b) The discrimination network distinguishes the generated carrier image and the real image, and changes the direction of gradient change aiming at the prediction result of the discrimination network;
c) And the generated network obtains the gradient returned by the discrimination network and updates the parameters to generate a new carrier image.
Further, the embedding process of the information in step 3) includes an embedding process using a fixed key and an embedding process using a random key.
Further, the embedding process using the fixed key adopts an LSB method for embedding; the embedding process using the random key adopts a HUGO method for embedding.
Further, the steganography analysis network is used for carrying out two-classification on the input carrier image and the steganography image, and the accuracy rate of classifying the carrier image and the steganography image into the original image and the steganography image is obtained.
Further, the steganalysis network is GNCNN.
Further, by setting different seed values, the deception of the generated carrier image to the steganalysis network after the information embedding is carried out under different parameters is compared, so that the information embedding safety is improved.
A secure steganography device based on a generative countermeasure network, comprising:
a generation network unit for generating a carrier image in which information is to be embedded under the generation countermeasure network framework;
the judging network unit is used for judging the authenticity of the generated carrier image and enabling the carrier image generated by the generating network unit to be close to a real image through a dynamic game process of the generating network unit;
and the steganography unit is used for embedding information into the carrier image generated by the network generation unit.
The steganography analysis network unit is used for carrying out two classifications on the input carrier image and the steganography image to obtain the accuracy rate of classifying the carrier image and the steganography image into the original image and the steganography image.
The method of the invention can be used for acting on the image embedded by the random key, and has the following advantages compared with the prior art:
1. the method provided by Denis and Burnaev (ICLR 2016Open review, 2016) is improved, and the problem that steganalysis effect of an image embedded by using a random key is not obvious in work is solved;
2. according to the invention, a WGAN (Wassertein genetic adaptive network, wassertein generation countermeasure network) is used for replacing a DCGAN, and the WGAN is used for generating an image, so that the generated image has better visual effect, is closer to a real image, reduces the training time and has higher training speed;
3. the discrimination network and the generation network are in mutual confrontation, so that the generated image is more beneficial to embedding, and the security of the steganographic image embedding is improved; the discrimination network is similar to the steganalysis network in structure but different in function, and the network and the generation network are in network confrontation, so that the generated image is more real and more suitable for embedding;
4. the steganography analysis network is adopted to evaluate whether the generated image is suitable for steganography, and experiments prove that the steganography analysis effect of the steganography image is better.
Drawings
FIG. 1 is a flow chart of image steganography and discrimination using the method of the present invention. The "data preprocessing" refers to performing a uniform cropping operation (e.g., performing a center cropping operation to uniformly crop the pictures into a size of 64 × 64) on the pictures.
Fig. 2 is an image generated by the generation network in the method of the present invention.
FIG. 3 is a flow chart of an image steganography embedding method using the method of the present invention, wherein (a) the graph uses a fixed key embedding method and (b) the graph uses a random key embedding method.
Detailed Description
The present invention will be described in further detail below with reference to specific examples and the accompanying drawings.
The steganography method based on the generation countermeasure network is suitable for embedding by using a random key and the same key, the flow of the method is shown in figure 1, and the method mainly comprises the following steps: firstly, a carrier image to be embedded with information is generated by a generating network, the authenticity of the generated carrier image is judged by a judging network of the WGAN, and through the dynamic game process of the generating network and the judging network, the carrier image closer to a real image is generated as much as possible to deceive the judging network, and the carrier image generated by the generating network is separated from the real image as much as possible by the target of the judging network. Then, information embedding (namely steganography) is carried out on the carrier image generated by the generation network, and the steganography analysis network is used for carrying out two classifications on the input image (comprising the carrier image and the steganography image) to obtain the accuracy of classifying the input image into the original image (namely the carrier image) and the steganography image (namely the steganography image).
The dynamic game process for generating the network and distinguishing the network comprises the following steps:
1. for a generation network, the data generated by the generation network is consistent with the real data as much as possible, namely the data are distributed in the same way, and a steganographic carrier image is generated;
2. the method comprises the steps that a carrier image and a real image which are generated by network learning are distinguished through judgment network, the direction of gradient change is changed according to a prediction result of a judgment network, and when the judgment network considers that the output of the generated network is real data and the output is noise data, the gradient updating direction of the judgment network is about to be changed; the gradient updating direction refers to the negative direction of the first derivative of the objective function;
3. and the generated network obtains the gradient returned by the judgment network, updates the parameters and generates a new carrier image. Due to the use of WGAN, the gradient disappearance problem of GAN does not occur and the network can be continuously updated. FIG. 2 is an exemplary diagram of an image generated by a generation network.
The process of embedding information by the method is carried out off line and is separated from the training process of the network. Firstly, embedding is carried out by using an LSB (Least Significant Bit) method, after a steganalysis result is obtained through a steganalysis network, embedding is carried out by using a HUGO (high undetected steGO) method again, and steganalysis is carried out through the steganalysis network. In the process of generating the image, different experimental parameters are set, and the degrees of network spoofing are analyzed by comparing the image generated under different parameters with steganography.
In the process of embedding information, the process of embedding LSB and HUGO twice is separately carried out, LSB is a fixed key embedding mode, HUGO is embedded by a random key, so that the method is proved to be effective for embedding the fixed key and the random key, and the effectiveness of the method is further proved. Fig. 3 illustrates the overall flow of the embedding method using both LSB and HUGO, where (a) the diagram uses a fixed key embedding method and (b) the diagram uses a random key embedding method.
The steganalysis Network In the method adopts GNCNN (Gaussian-Neuron Convolutional Neural Network), which IS referred to as InIS & T/SPIE Electronic Imaging, pp.94090J-94090J. GNCNN proposes a new approach to steganalysis by automatically learning features through a deep learning model that can capture complex dependencies of features useful for steganalysis. Compared with the existing steganalysis mechanism, the model can automatically learn feature representation by using several convolution layers. The steps of feature extraction and classification are unified under the same architecture, which means that classification can be used for guidance in the process of feature extraction.
Example 1 secure steganography method based on generation of countermeasure network
Take the CelebA face data set as an example:
1) Inputting a noise signal z to a generating network G under a WGAN (Wasserstein generated adaptive Networks, wasserstein generation countermeasure network) framework, and generating a carrier image I by the generating network G;
2) Judging the mutual game between the network D and the generating network G, so that the carrier image I generated by G is closer to a real image;
3) The discrimination network D discriminates and analyzes the carrier image I generated by the G to obtain the probability of judging the carrier image I into a true image and a false image;
4) Embedding the generated carrier image I into information in an off-line manner, firstly embedding the information by using an LSB fixed key to obtain an image I 'subjected to steganography, simultaneously inputting the I and the I' into a steganography analysis network S, and judging the probability of an original image or a steganography;
5) Then, using HUGO to randomly embed to obtain a steganographic image I ', simultaneously inputting I and I' into a steganographic analysis network S, and judging the probability of an original image or a steganographic image;
6) Repeating the steps 1) to 5) under a DCGAN (Deep convolution generated confrontation network) framework, and classifying the original image and the steganographic image through a steganographic analysis network to obtain a probability;
7) The classification accuracy obtained by comparing the probabilities under the two methods is shown in table 1:
TABLE 1 Classification accuracy by training steganalysis network on real images
| Image type | The method of the invention | SGANs |
| Original drawing | 0.87 | 0.92 |
| Generated byImage of a person | 0.72 | 0.90 |
In table 1, SGANs is a comparative method, which is implemented using DCGAN; "generated image" refers to a carrier image generated by generating a network. In the experiment, steganography is respectively performed on an original image and a generated carrier image, and then steganography analysis is performed to obtain the classification accuracy. After a plurality of training tests, the method provided by the invention has lower classification accuracy of the generated images, namely, the generated images are not easy to be distinguished from real images, so that the safety is higher than that of a comparison method.
Example 2 secure steganography method based on generation of countermeasure network
Taking the CelebA face data set as an example, in the experiment, different seed values are set to compare deception of the generated image on the steganalysis network after the information is embedded under different parameters. The seed value is used for controlling the repeatability of the experiment under the condition of a certain random number, and is embodied as controlling the randomness of the generated image in the invention.
1) Under the WGAN framework, inputting a noise signal z into a generating network G, and generating a carrier image I by the generating network G;
2) Judging the mutual game of the network D and the generating network G, and generating a carrier image I which is closer to a real image;
3) The discrimination network D discriminates and analyzes the generated carrier image I to obtain the probability of judging the carrier image I into a true image and a false image;
4) Using the same seed value, embedding information into a generated carrier image I by using an LSB method to obtain an image I 'subjected to steganography, simultaneously inputting the I and the I' into a steganography analysis network S, and judging the probability of an original image or a steganography;
5) Then, using HUGO to randomly embed to obtain an image I after steganography, simultaneously inputting the I and the I' into a steganography analysis network S, and judging the probability of the original image or the steganography;
6) Embedding information into the generated carrier image I by using a randomly generated seed value to obtain an image T after steganography, simultaneously inputting the I and the T into a steganography analysis network S, and judging the probability of the original image or the steganography;
7) Using a randomly generated seed value, finely adjusting the WGAN in the training process, adjusting the learning rate, the Momentum coefficient and the like to generate an image embedded information, obtaining an image M after steganography, simultaneously inputting the I and the M into a steganography analysis network S, and judging the probability of an original image or a steganography;
8) Classification accuracy was compared at three experimental settings, as shown in table 2:
TABLE 2 Classification accuracy rates obtained by steganalysis network training on generated images under different experimental settings
| Conditions of the experiment | Accuracy of classification |
| 4) | 0.87 |
| 6) | 0.72 |
| 7) | 0.71 |
In table 2, 4), 6), 7) represent the experimental setup in steps 4), 6), 7), respectively. Experimental results show that in the process of generating the image, different seed values are set, so that lower classification accuracy can be obtained, namely, the steganalysis network can be deceived, and therefore the image generated by the method is more suitable for steganography and higher in safety.
Example 3 secure steganography device based on generation of countermeasure network
The security steganography device based on the generation countermeasure network comprises: a generation network unit for generating a carrier image in which information is to be embedded under the generation countermeasure network framework; the judging network unit is used for judging the authenticity of the generated carrier image and enabling the carrier image generated by the generating network unit to be close to a real image through a dynamic game process of the generating network unit; and the steganography unit is used for embedding information into the carrier image generated by the network generation unit. Further, the device also comprises a steganography analysis network unit which is used for carrying out two classifications on the input carrier image and the steganography image to obtain the accuracy rate of classifying the carrier image and the steganography image into an original image and a steganography image.
The invention is carried out under the framework of WGAN (Wasserstein generated countermeasure network). Aiming at the visual quality of the generated image, the generation network and the discrimination network can be replaced by other generation countermeasure networks with high quality and high generation speed of the generated image; aiming at the steganography method, the invention carries out experiments of fixed key embedding and random key embedding, embodies the effectiveness of the steganography method provided by the invention through the experimental results of the two methods, and can also adopt other self-adaptive steganography to carry out experiments; aiming at the steganalysis method, the GNCNN network is preferably used in the invention, and the steganalysis detection rate is higher.
The above embodiments are only intended to illustrate the technical solution of the present invention and not to limit the same, and a person skilled in the art can modify the technical solution of the present invention or substitute the same without departing from the spirit and scope of the present invention, and the scope of the present invention should be determined by the claims.
Claims (3)
1. A security steganography method based on a generation countermeasure network is characterized by comprising the following steps:
1) Under the framework of generating a countermeasure network, generating a carrier image in which information is to be embedded through a generation network, and judging the authenticity of the generated carrier image through a judgment network; the generative countermeasure network is a WGAN;
2) Through the dynamic game processes of generating a network and judging the network, the carrier image generated by the generated network is close to a real image; the dynamic game process for generating the network and judging the network comprises the following steps:
a) For the generation network, the data it produces is made to be as consistent as possible with the actual data, thus generating the carrier image in which the information is to be embedded;
b) The discrimination network distinguishes the generated carrier image and the real image, and changes the direction of gradient change aiming at the prediction result of the discrimination network;
c) The generation network obtains the gradient returned by the discrimination network and updates the parameters to generate a new carrier image;
3) Embedding information into a carrier image generated by a generation network, and performing secondary classification on the input carrier image and the steganographically-generated image by using a steganographic analysis network to obtain the accuracy rate of classifying the carrier image into an original image and a steganographically; different seed values are set to compare the deception of the generated carrier image to the steganalysis network after the information is embedded under different parameters, so that the information embedding safety is improved;
wherein, the step 3) of embedding information into the carrier image generated by the generation network comprises the following steps:
embedding by using an LSB (least significant bit) method, obtaining a steganalysis result through a steganalysis network, then embedding by using a HUGO method again, and performing steganalysis through the steganalysis network;
in the process of embedding information, the process of embedding twice by using the LSB and the HUGO is separately carried out, wherein the LSB is embedded by adopting a fixed key, and the HUGO is embedded by adopting a random key;
the steganalysis network is used for evaluating whether the generated image is suitable for steganography.
2. The method of claim 1, wherein the steganalysis network is a GNCNN.
3. A secure steganographic apparatus based on a generative countermeasure network employing the method of claim 1, comprising:
a generation network unit for generating a carrier image in which information is to be embedded under the generation countermeasure network framework;
the judging network unit is used for judging the authenticity of the generated carrier image and enabling the carrier image generated by the generating network unit to be close to a real image through a dynamic game process of the generating network unit;
the steganography unit is used for embedding information into the carrier image generated by the network generation unit;
and the steganography analysis network unit is used for carrying out two classifications on the input carrier image and the steganography image to obtain the accuracy rate of classifying the carrier image and the steganography image into the original image and the steganography image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710670786.1A CN107563155B (en) | 2017-08-08 | 2017-08-08 | Security steganography method and device based on generation of countermeasure network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710670786.1A CN107563155B (en) | 2017-08-08 | 2017-08-08 | Security steganography method and device based on generation of countermeasure network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107563155A CN107563155A (en) | 2018-01-09 |
| CN107563155B true CN107563155B (en) | 2023-02-28 |
Family
ID=60973962
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710670786.1A Active CN107563155B (en) | 2017-08-08 | 2017-08-08 | Security steganography method and device based on generation of countermeasure network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107563155B (en) |
Families Citing this family (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108154547B (en) * | 2018-01-17 | 2019-08-09 | 百度在线网络技术（北京）有限公司 | Image generating method and device |
| CN108346125B (en) * | 2018-03-15 | 2021-10-08 | 中山大学 | Airspace image steganography method and system based on generation countermeasure network |
| CN108765246B (en) * | 2018-04-03 | 2019-07-16 | 宁波大学 | A kind of selection method of steganographic system carrier image |
| CN109002686B (en) * | 2018-04-26 | 2022-04-08 | 浙江工业大学 | Multi-grade chemical process soft measurement modeling method capable of automatically generating samples |
| CN110457910A (en) * | 2018-05-07 | 2019-11-15 | 中国人民武装警察部队工程大学 | A kind of production information concealing method based on image synthesis |
| CN108596141B (en) * | 2018-05-08 | 2022-05-17 | 深圳大学 | Detection method and system for generating face image by deep network |
| CN108648135B (en) * | 2018-06-01 | 2022-05-27 | 深圳大学 | Hidden model training and using method, device and computer readable storage medium |
| CN109348211B (en) * | 2018-08-06 | 2020-11-06 | 中国科学院声学研究所 | General information hiding detection method for video intra-frame inter-frame coding |
| CN109214973B (en) * | 2018-08-24 | 2020-10-27 | 中国科学技术大学 | Method for generating countermeasure security carrier aiming at steganalysis neural network |
| CN109359667A (en) * | 2018-09-07 | 2019-02-19 | 华南理工大学 | A kind of feature recalibration convolution method based on WGAN model |
| CN109993678B (en) * | 2019-03-26 | 2020-04-07 | 南京联创北斗技术应用研究院有限公司 | Robust information hiding method based on deep confrontation generation network |
| CN110084734B (en) * | 2019-04-25 | 2023-02-14 | 南京信息工程大学 | Big data ownership protection method based on object local generation countermeasure network |
| CN110334805B (en) * | 2019-05-05 | 2022-10-25 | 中山大学 | JPEG domain image steganography method and system based on generation countermeasure network |
| CN111768325B (en) * | 2020-04-03 | 2023-07-25 | 南京信息工程大学 | Security improvement method based on generation of countermeasure sample in big data privacy protection |
| CN111768326B (en) * | 2020-04-03 | 2023-08-25 | 南京信息工程大学 | High-capacity data protection method based on GAN (gas-insulated gate bipolar transistor) amplified image foreground object |
| CN112115490B (en) * | 2020-08-14 | 2023-09-26 | 石坚 | GAN-based carrier image synthesis steganography method |
| CN112395635B (en) * | 2021-01-18 | 2021-05-04 | 北京灵汐科技有限公司 | Image processing method, device, secret key generating method, device, training method and device, and computer readable medium |
| CN114339258B (en) * | 2021-12-28 | 2024-05-10 | 中国人民武装警察部队工程大学 | Information steganography method and device based on video carrier |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106920206A (en) * | 2017-03-16 | 2017-07-04 | 广州大学 | A kind of steganalysis method based on confrontation neutral net |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105979268A (en) * | 2016-05-05 | 2016-09-28 | 北京智捷伟讯科技有限公司 | Safe information transmission method based on lossless watermark embedding and safe video hiding |
-
2017
- 2017-08-08 CN CN201710670786.1A patent/CN107563155B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106920206A (en) * | 2017-03-16 | 2017-07-04 | 广州大学 | A kind of steganalysis method based on confrontation neutral net |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107563155A (en) | 2018-01-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107563155B (en) | Security steganography method and device based on generation of countermeasure network | |
| Zhang et al. | Protecting intellectual property of deep neural networks with watermarking | |
| Li et al. | Concealed attack for robust watermarking based on generative model and perceptual loss | |
| Avcıbaş et al. | Image steganalysis with binary similarity measures | |
| Wang et al. | Data hiding with deep learning: A survey unifying digital watermarking and steganography | |
| CN107240061B (en) | Watermark embedding and extracting method and device based on dynamic BP neural network | |
| CN108717680B (en) | Airspace image steganalysis method based on full-dense connection network | |
| Zhou et al. | On security enhancement of steganography via generative adversarial image | |
| Geetha et al. | Optimized image steganalysis through feature selection using MBEGA | |
| CN109344578A (en) | Based on the insertion of the audio frequency watermark of chaos and wavelet transformation, extracting method | |
| CN112561770A (en) | Confrontation sample defense method based on fragile watermark | |
| Meng et al. | High-capacity steganography using object addition-based cover enhancement for secure communication in networks | |
| Gupta et al. | A survey of watermarking technique using deep neural network architecture | |
| Huang et al. | Block-based adaptive image steganography using LSB matching revisited | |
| CN117176874A (en) | Embedded-image-free steganography method and device based on bidirectional generation countermeasure network | |
| WO2023142453A1 (en) | Biometric identification method, server, and client | |
| Shen et al. | IrisMarkNet: Iris feature watermarking embedding and extraction network for image copyright protection | |
| Yongqiang et al. | A DWT domain image watermarking scheme using genetic algorithm and synergetic neural network | |
| CN115546003A (en) | Back door watermark image data set generation method based on confrontation training network | |
| Ramezani et al. | Adaptive image steganography with mod-4 embedding using image contrast | |
| Loukhaoukha et al. | Multi-Objective genetic algorithm optimization for image watermarking based on singular value decomposition and Lifting wavelet transform | |
| Desai et al. | Survey on universal image steganalysis | |
| Zhang et al. | Technology of image steganography and steganalysis based on adversarial training | |
| Tian et al. | Analysis of electronic bill authentication and security storage performance utilising machine learning algorithm | |
| Abolghasemi et al. | Steganalysis of LSB matching based on co-occurrence matrix and removing most significant bit planes |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |