CN109587372A - A kind of invisible image latent writing art based on generation confrontation network - Google Patents
A kind of invisible image latent writing art based on generation confrontation network Download PDFInfo
- Publication number
- CN109587372A CN109587372A CN201811509096.9A CN201811509096A CN109587372A CN 109587372 A CN109587372 A CN 109587372A CN 201811509096 A CN201811509096 A CN 201811509096A CN 109587372 A CN109587372 A CN 109587372A
- Authority
- CN
- China
- Prior art keywords
- image
- network
- close
- secret
- latent writing
- 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.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
- H04N1/32—Circuits or arrangements for control or supervision between transmitter and receiver or between image input and image output device, e.g. between a still-image camera and its memory or between a still-image camera and a printer device
- H04N1/32101—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title
- H04N1/32144—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title embedded in the image data, i.e. enclosed or integrated in the image, e.g. watermark, super-imposed logo or stamp
- H04N1/32149—Methods relating to embedding, encoding, decoding, detection or retrieval operations
- H04N1/32309—Methods relating to embedding, encoding, decoding, detection or retrieval operations in colour image data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
- Editing Of Facsimile Originals (AREA)
Abstract
The present invention discloses a kind of invisible image latent writing art based on generation confrontation network, can be realized one gray scale Secret Image of insertion in a color host image and obtains carrying close image, and can successfully recover Secret Image from the close image of load.Include: encoder network, is responsible for being embedded into Secret Image to generate in carrier image carrying close image；Decoder network is responsible for recovering Secret Image from the close image of load；Arbiter network is responsible for natural image and carries close image progress steganalysis, to adjust the safety of encoder network and decoder network.The present invention, based on the image latent writing art for generating confrontation network, is hidden for image information by construction and provides new mentality of designing.
Description
Technical field
The invention belongs to field of image processings, more particularly to based on the image latent writing art for generating confrontation network.
Background technique
Recently as internet fast development with it is universal, communicating has become more and more convenient, but to information security
It also proposed new challenge.On the one hand, people always have some secret informations to be not intended to be learned by third party in the communications；Another party
Face, emerging digital multimedia works need copyright protection, and there are also a large amount of electronic commerce datas etc. to need integrality true
Recognize.Traditional cryptography not can be well solved emerging these problems.Steganography (Steganography) is used for
Secret information is embedded into normal carrier without the perception characteristics for changing carrier, by carry the transmission of close carrier in the channel come
Realize the transmission of secret information.Steganography not only conceals the content of communication, also conceals the behavior of communication, therefore can be fine
It applies in scene as above on ground.Digital picture has become one of data type most frequently used on internet, has
The big feature of capacity of big, the embeddable secret information of redundancy, therefore the Steganography of mainstream is using image as carrier mostly at present
Image latent writing art (Image Steganography).
There are three the index of measurement image latent writing art is main: capacity, invisibility, safety.Capacity refers to embeddable number
The length of secret information in word image；Invisibility refers to whether carrier image can be with original graph after secret information insertion
As indifference；Whether safety then requires to carry close image and can break away from being attacked in transmission process the detection of algorithm.Existing figure
As all more or less there are some problems at this three aspect for Steganography.
On the other hand, with the continuous development of image latent writing art, for the detection algorithm of image latent writing art --- image is hidden
Analysis (Image Steganalysis) algorithm is write also to be evolving.The purpose of image latent writing analysis is by carrying out to image
Analysis is to judge whether contain secret information in image, it might even be possible to which estimated information embedded quantity obtains secret information, destroys secret
Information etc..For existing image latent writing art, constantly there is new image latent writing parser to be suggested, and performance is increasingly
It is good.It is therefore proposed that new image latent writing art effectively hides steganalysis to meet the needs of more and more application scenarios
The detection of algorithm, is of great importance for the network information security.
Existing image latent writing art often relies on well-designed algorithm, carries out in the airspace of image or transform domain secret
The insertion of information.This kind of Steganographies need a large amount of priori knowledge, and algorithm once designs and just cannot occur further according to new
Steganalysis algorithm be automatically adjusted.Therefore designing the new image latent writing art of one kind is a huge challenge.
Recently as the arrival of computing capability being substantially improved with big data era, using convolutional neural networks as representative
Deep learning method achieves excellent effect in the tasks such as image recognition, target detection, image generation.Deep learning passes through
Convolutional neural networks to image information carry out feature extraction with merge, and have supervise or semi-supervised ground by way of realize that convolution is refreshing
Parameter through network updates, and is finally completed some specific task.The newest fruits that confrontation network is deep learning field are generated,
It includes generators and arbiter two parts.By the dual training of generator and arbiter, generating confrontation network can be generated
The sample for thering is set of metadata of similar data to be distributed with authentic specimen.Confrontation network is generated to appoint in image generation, Style Transfer, speech synthesis etc.
Excellent effect is achieved in business.
Summary of the invention
The present invention propose it is a kind of based on the digital picture steganographic algorithm for generating confrontation network, by convolutional neural networks from
Dynamic study is final to realize the invisible Steganography that a gray scale Secret Image is embedded in a color host image, and safety
It is continuously improved in the learning process of neural network.
The present invention provides a kind of based on the image latent writing method for generating confrontation network, comprising the following steps:
1) under coder-decoder network frame, by encoder network by Secret Image steganography into carrier image
It generates and carries close image, recovered Secret Image from the close image of load by decoder network；
2) in the case where generating confrontation network frame, encoder network and decoder network form a model conduct end to end
Generator network generates the Secret Image for carrying close image and recovering by generator network, by arbiter network to generation
The true and false property of the close image of load judged；
3) pass through the dual training process of generator network and arbiter network, so that the close figure of load that generator network generates
As consistent with corresponding carrier image and close to true picture so that the Secret Image that goes out of generator network recovery with
Original private image is similar as much as possible.
4) using the image training model of sizes size, so that model has better generalization ability.
Further, the dual training process of the generator network and arbiter network includes:
A) for generator network, encoder network therein will make the close image of load generated and carrier image as much as possible
Similar, decoder network therein will make the Secret Image recovered and original private image similar as much as possible；
B) arbiter network plays a part of steganalysis, differentiates that input picture is that natural image or generator network are raw
At the close image of load, for steganalysis as a result, the direction to change of gradient is changed；
C) generator network obtains the parameter of gradient updating encoder network and decoder network that arbiter network is passed back,
Generation preferably carries close image.
Further, gray scale Secret Image is only embedded into the YCrCb color of carrier image by encoder network in step 1)
In the channel Y under color space, decoder network also only recovers from the channel Y under the YCrCb color space for carrying close image secret
Close image.
Further, different characteristic can be fused to by the encoder network in step 1) by Inception module composition
The steganography of image is realized together；Decoder network is a full convolutional coding structure.The two can handle the image of arbitrary dimension.
Further, the arbiter in step 2) is the steganalysis model realized by convolutional neural networks, wherein
The characteristic pattern of arbitrary dimension is mapped to the feature vector fixed to regular length by use space pyramid pond module, is broken
Size limitation to input picture, has better steganalysis performance.
Further, it will use a recombination losses function in step 3) to instruct the training of generator network.
Secret Image can be embedded into carrier image well using method of the invention, had compared with prior art
It has the advantage that
1, the present invention improves the method for the propositions such as Baluja etc. and Atique, solves and carries close image and carrier
Image discrepant problem on color；
2, the present invention is trained using confrontation network is generated, and makes model can be according to hidden during realizing image latent writing
The analysis situation for writing analyzer carrys out adjusting parameter, and the close image of the load of generation is less susceptible to be detected by steganalysis device；
3, the present invention carrys out the training of guide image steganography process using a recombination losses function, accelerates training speed,
Improve the quality for carrying close image and the Secret Image recovered.
Detailed description of the invention
Fig. 1 is the flow chart that the method for the present invention carries out image latent writing.Wherein, encoder network and decoder network constitute life
It grows up to be a useful person network, steganalysis device is as arbiter network of the invention.
Fig. 2 is the instance graph that image latent writing is carried out using the method for the present invention.
Fig. 3 is the method for the present invention in different frequency of training, the recall rate contrast table of steganalysis model.
Specific embodiment
To be clearer and more comprehensible These characteristics and advantage of the invention, With reference to embodiment with attached drawing to this hair
It is bright to be described in further detail.
The image latent writing method that the present invention designs is based on generating confrontation network, suitable for gray scale Secret Image is embedded into coloured silk
In color carrier image.This method is trained to obtain optimal model parameters by using data the set pair analysis model, specifically trains process
As shown in Figure 1, its key step includes:
Color host image is transformed into YCrCb color space by step 101 by rgb color space.
Step 102, by the channel Y of color host image and gray scale Secret Image splicing (concatenate) to together it
After be output in encoder network.
Step 103, encoder network are made by feature extraction with one single pass image of output later, the image is merged
To carry the channel Y of close image and two combination of channels of Cr, Cb of initial carrier image to rgb color space is transformed into together, obtain
Colour carries close image.
Step 201 will carry close image and be transformed into YCrCb color space, and the channel Y therein is only input to decoder
In network.
Step 202, decoder network by feature extraction with merge export a single channel image, what is as recovered is secret
Close image.
Step 203, calculated using designed recombination losses function carry difference between close image and initial carrier image,
Difference between the Secret Image recovered and original private image, will be both as a part of loss function value.
Step 301 carries out steganalysis to physical carrier image pattern using arbiter network, judges steganalysis result
It is whether consistent with true classification；
Step 302 carries out steganalysis using the close image of load of the arbiter network to generation, judge steganalysis result with
Whether true classification is consistent；
Step 401 calculates the difference carried between close image pattern and physical carrier image pattern using recombination losses function
Difference value between the Secret Image be worth, recovered and original private image, the differentiation according to arbiter network to close image is carried
As a result differentiation penalty values are obtained, three is added to the damage as generator network (encoder network and decoder network) together
It loses, calculates gradient, update the parameter of generator network (encoder network and decoder network)；
The penalty values of step 402, the steganalysis result computational discrimination device obtained according to step 3, calculate gradient, and update is sentenced
The parameter of other device network.
Step 501 completes the wheel on training set after training, and verification test is carried out on test set, it is close to calculate load
The average similarity of the average similarity of image and carrier image, the Secret Image recovered and original private image；
Step 502 checks whether the result of step 501 reaches expectation index, if having carry out next step operation, it is no then
Return to the training that step A starts next round on training set；
Step 503 is finely adjusted training to model using the image pattern of sizes, improves the generalization ability of model.
The present invention is trained and has been verified on LFW, Pascal VOC2012 and tri- data sets of ImageNet, Fig. 2
In be by the method for the present invention carry out image latent writing sample instantiation figure, it can be seen that carry close image and initial carrier picture altitude one
It causes, and the Secret Image recovered is consistent with original private picture altitude.It is the method for the present invention in Fig. 3 in different frequency of training
When, steganalysis algorithm compares the recall rate for carrying close image.This comparison is shown with the continuous instruction for generating confrontation network
Practice, the safety of the method for the present invention is being continuously improved.
The present invention realizes one gray scale secret figure of insertion in a color image under the frame of coder-decoder
The image latent writing task that Secret Image recovers simultaneously can be allowed steganography model instructing by picture under the frame for generating confrontation network
The attack of steganalysis algorithm is considered during practicing, continuous adjusting parameter is to improve the safety of model.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
Within the technical scope disclosed by the invention, any changes or substitutions that can be easily thought of by any people for being familiar with the technology, should all cover
Within protection scope of the present invention.Therefore, protection scope of the present invention should be subject to the protection scope in claims.
Claims (6)
1. a kind of based on the invisible image latent writing art for generating confrontation network characterized by comprising
A, generate and carry close image: encoder network is realized by convolutional neural networks, is responsible for gray scale Secret Image being embedded into colour
In carrier image, the colored close image of load is generated；
B, recover Secret Image: decoder network is realized by full convolutional neural networks, is responsible for carrying in close image from colour and be restored
Gray scale Secret Image out；
C, carry out steganalysis using arbiter network: arbiter network generates the natural image or encoder network of input
It carries close image and carries out steganalysis；
D, parameter update and repetitive exercise: penalty values are calculated using coincidence loss function, calculate gradient, undated parameter；
E, verifying model performance and extensive training: by structural similarity index verification model performance, using multiple dimensioned sample into
The extensive training of row.
2. according to claim 1 a kind of based on the invisible image latent writing art for generating confrontation network, which is characterized in that step
Rapid A is further included steps of
A1, color host image are transformed into YCrCb color space by rgb color space；
A2, it is input in encoder network, passes through after the channel Y of color host image and gray scale Secret Image are spliced together
It crosses the feature extraction of encoder network and generates a single channel image with merging.
Two combination of channels of Cr, Cb of A3, the single channel image that previous step is generated and initial carrier image are final to constituting together
Colour carry close image.
3. according to claim 1 a kind of based on the invisible image latent writing art for generating confrontation network, which is characterized in that step
Rapid B is further included steps of
B1, the close image of load that step A is generated is transformed into YCrCb color space, only the channel Y is input in decoder network
B2, decoder network by feature extraction with merge export a single channel image, the Secret Image as recovered.
4. according to claim 1 a kind of based on the invisible image latent writing art for generating confrontation network, which is characterized in that step
It is further included steps of under rapid C
C1, physical carrier image pattern is inputted into arbiter network, obtain analysis classification, determine whether and true classification one
It causes；
C2, into arbiter network input step A generate the close image pattern of load, obtain analysis classification, determine whether with really
Classification is consistent.
5. according to claim 1 a kind of based on the invisible image latent writing art for generating confrontation network, which is characterized in that step
Rapid D is further included steps of
D1, it is calculated using recombination losses function and carries difference value between close image pattern and physical carrier image pattern, recovers
Secret Image and original private image between difference value, the differentiation result for carrying close image is obtained according to arbiter network and is sentenced
Three is added to the loss as generator network (encoder network and decoder network) together, calculates ladder by other penalty values
Degree updates the parameter of generator network (encoder network and decoder network)；
D2, the penalty values according to the result computational discrimination device network of step C, calculate gradient, update the parameter of arbiter network.
6. according to claim 1 a kind of based on the invisible image latent writing art for generating confrontation network, which is characterized in that step
Rapid E is further included steps of
E1, it completes the wheel on training set after training, verification test is carried out on test set, calculates and carries close image and carrier
The average similarity of the average similarity of image, the Secret Image recovered and original private image；
E2, it checks whether the result of step E1 reaches expectation index, next step operation is carried out if having, does not return to step A then
Start the training of next round on training set；
E3, training is finely adjusted to model using the image pattern of sizes, improves the generalization ability of model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811509096.9A CN109587372B (en) | 2018-12-11 | 2018-12-11 | Invisible image steganography based on generation of countermeasure network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811509096.9A CN109587372B (en) | 2018-12-11 | 2018-12-11 | Invisible image steganography based on generation of countermeasure network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109587372A true CN109587372A (en) | 2019-04-05 |
| CN109587372B CN109587372B (en) | 2020-06-16 |
Family
ID=65928897
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811509096.9A Active CN109587372B (en) | 2018-12-11 | 2018-12-11 | Invisible image steganography based on generation of countermeasure network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109587372B (en) |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110334805A (en) * | 2019-05-05 | 2019-10-15 | 中山大学 | A kind of JPEG domain image latent writing method and system based on generation confrontation network |
| CN110502646A (en) * | 2019-08-02 | 2019-11-26 | 广州多益网络股份有限公司 | A kind of multimedia file purification method, device, storage medium and terminal device |
| CN110503157A (en) * | 2019-08-27 | 2019-11-26 | 安徽大学 | The image latent writing analysis method of multitask convolutional neural networks based on fine granularity image |
| CN110889797A (en) * | 2019-10-15 | 2020-03-17 | 浙江大学 | Depth self-adaptive image hiding method based on confrontation sample generation |
| CN111598762A (en) * | 2020-04-21 | 2020-08-28 | 中山大学 | Generating type robust image steganography method |
| CN111681154A (en) * | 2020-06-09 | 2020-09-18 | 湖南大学 | Color image steganography distortion function design method based on generation countermeasure network |
| CN111768325A (en) * | 2020-04-03 | 2020-10-13 | 南京信息工程大学 | Security improvement method based on generation of countermeasure sample in big data privacy protection |
| CN112243005A (en) * | 2020-10-14 | 2021-01-19 | 合肥工业大学 | Secure non-embedded steganography method based on generation of countermeasure network |
| CN112561773A (en) * | 2020-12-28 | 2021-03-26 | 中国科学技术大学 | Deep disturbance-based anti-steganography method |
| CN112884630A (en) * | 2021-02-04 | 2021-06-01 | 南开大学 | Reversible neural network-based high-capacity image steganography and recovery method |
| CN113077377A (en) * | 2021-05-13 | 2021-07-06 | 海南大学 | Color image steganography method based on generation countermeasure network |
| CN113099066A (en) * | 2019-12-23 | 2021-07-09 | 浙江工商大学 | Large-capacity image steganography method based on multi-scale fusion cavity convolution residual error network |
| CN113254891A (en) * | 2021-05-17 | 2021-08-13 | 山东大学 | Information hiding method, device and system based on void space pyramid |
| CN113326531A (en) * | 2021-06-29 | 2021-08-31 | 湖南汇视威智能科技有限公司 | Robust efficient distributed face image steganography method |
| CN113538202A (en) * | 2021-08-05 | 2021-10-22 | 齐鲁工业大学 | Image steganography method and system based on generative steganography confrontation |
| CN113989097A (en) * | 2021-12-30 | 2022-01-28 | 南京中孚信息技术有限公司 | Information steganography model training method, information steganography device and storage medium |
| CN114025053A (en) * | 2021-09-28 | 2022-02-08 | 上海电力大学 | Image steganography method combining high color similarity preprocessing |
| CN114207659A (en) * | 2019-06-12 | 2022-03-18 | 罗格斯，新泽西州立大学 | Light field messaging |
| CN114220443A (en) * | 2021-11-04 | 2022-03-22 | 合肥工业大学 | BN optimization SNGAN-based training method and system for adaptive audio steganography model |
| WO2022127374A1 (en) * | 2020-12-18 | 2022-06-23 | 海南大学 | Color image steganography method based on convolutional neural network |
| CN115022649A (en) * | 2022-05-17 | 2022-09-06 | 薇链信息技术有限公司 | Reversible neural network-based high-embedding-quantity and colorizable image steganography and recovery method |
| CN115086674A (en) * | 2022-06-16 | 2022-09-20 | 西安电子科技大学 | Image steganography method based on generation of countermeasure network |
| CN117132671A (en) * | 2023-10-27 | 2023-11-28 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Multi-task steganography method, system and medium based on depth self-adaptive steganography network |
| CN117292696A (en) * | 2023-10-08 | 2023-12-26 | 合肥工业大学 | End-to-end audio steganography method, system, storage medium and electronic equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108171266A (en) * | 2017-12-25 | 2018-06-15 | 中国矿业大学 | A kind of learning method of multiple target depth convolution production confrontation network model |
| CN108346125A (en) * | 2018-03-15 | 2018-07-31 | 中山大学 | A kind of spatial domain picture steganography method and system based on generation confrontation network |
| CN108921764A (en) * | 2018-03-15 | 2018-11-30 | 中山大学 | A kind of image latent writing method and system based on generation confrontation network |
| CN108923922A (en) * | 2018-07-26 | 2018-11-30 | 北京工商大学 | A kind of text steganography method based on generation confrontation network |
-
2018
- 2018-12-11 CN CN201811509096.9A patent/CN109587372B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108171266A (en) * | 2017-12-25 | 2018-06-15 | 中国矿业大学 | A kind of learning method of multiple target depth convolution production confrontation network model |
| CN108346125A (en) * | 2018-03-15 | 2018-07-31 | 中山大学 | A kind of spatial domain picture steganography method and system based on generation confrontation network |
| CN108921764A (en) * | 2018-03-15 | 2018-11-30 | 中山大学 | A kind of image latent writing method and system based on generation confrontation network |
| CN108923922A (en) * | 2018-07-26 | 2018-11-30 | 北京工商大学 | A kind of text steganography method based on generation confrontation network |
Non-Patent Citations (3)
| Title |
|---|
| ATIQUE UR REHMAN，ET AL: "END-TO-END TRAINED CNN ENCODER-DECODER NETWORKS FOR IMAGE STEGANOGRAPHY", 《ARXIV.ORG》 * |
| SHUMEET BALUJA: "Hiding Images in Plain Sight: Deep Steganography", 《31ST CONFERENCE ON NEURAL INFORMATION PROCESSING SYSTEMS (NIPS 2017)》 * |
| 翟黎明 等: "深度学习在图像隐写术与隐写分析领域中的研究进展", 《信息安全学报》 * |
Cited By (39)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110334805B (en) * | 2019-05-05 | 2022-10-25 | 中山大学 | JPEG domain image steganography method and system based on generation countermeasure network |
| CN110334805A (en) * | 2019-05-05 | 2019-10-15 | 中山大学 | A kind of JPEG domain image latent writing method and system based on generation confrontation network |
| CN114207659A (en) * | 2019-06-12 | 2022-03-18 | 罗格斯，新泽西州立大学 | Light field messaging |
| CN110502646A (en) * | 2019-08-02 | 2019-11-26 | 广州多益网络股份有限公司 | A kind of multimedia file purification method, device, storage medium and terminal device |
| CN110503157A (en) * | 2019-08-27 | 2019-11-26 | 安徽大学 | The image latent writing analysis method of multitask convolutional neural networks based on fine granularity image |
| CN110503157B (en) * | 2019-08-27 | 2021-11-19 | 安徽大学 | Image steganalysis method of multitask convolution neural network based on fine-grained image |
| CN110889797B (en) * | 2019-10-15 | 2021-06-08 | 浙江大学 | Depth self-adaptive image hiding method based on confrontation sample generation |
| CN110889797A (en) * | 2019-10-15 | 2020-03-17 | 浙江大学 | Depth self-adaptive image hiding method based on confrontation sample generation |
| CN113099066A (en) * | 2019-12-23 | 2021-07-09 | 浙江工商大学 | Large-capacity image steganography method based on multi-scale fusion cavity convolution residual error network |
| CN113099066B (en) * | 2019-12-23 | 2022-09-30 | 浙江工商大学 | Large-capacity image steganography method based on multi-scale fusion cavity convolution residual error network |
| CN111768325A (en) * | 2020-04-03 | 2020-10-13 | 南京信息工程大学 | Security improvement method based on generation of countermeasure sample in big data privacy protection |
| CN111598762A (en) * | 2020-04-21 | 2020-08-28 | 中山大学 | Generating type robust image steganography method |
| CN111598762B (en) * | 2020-04-21 | 2023-09-19 | 中山大学 | Generating type robust image steganography method |
| CN111681154B (en) * | 2020-06-09 | 2022-06-21 | 湖南大学 | Color image steganography distortion function design method based on generation countermeasure network |
| CN111681154A (en) * | 2020-06-09 | 2020-09-18 | 湖南大学 | Color image steganography distortion function design method based on generation countermeasure network |
| CN112243005A (en) * | 2020-10-14 | 2021-01-19 | 合肥工业大学 | Secure non-embedded steganography method based on generation of countermeasure network |
| WO2022127374A1 (en) * | 2020-12-18 | 2022-06-23 | 海南大学 | Color image steganography method based on convolutional neural network |
| CN112561773B (en) * | 2020-12-28 | 2024-03-29 | 中国科学技术大学 | Deep disturbance-based countersteganography method |
| CN112561773A (en) * | 2020-12-28 | 2021-03-26 | 中国科学技术大学 | Deep disturbance-based anti-steganography method |
| CN112884630A (en) * | 2021-02-04 | 2021-06-01 | 南开大学 | Reversible neural network-based high-capacity image steganography and recovery method |
| CN112884630B (en) * | 2021-02-04 | 2022-06-28 | 南开大学 | Reversible neural network-based high-capacity image steganography and recovery method |
| CN113077377A (en) * | 2021-05-13 | 2021-07-06 | 海南大学 | Color image steganography method based on generation countermeasure network |
| CN113254891A (en) * | 2021-05-17 | 2021-08-13 | 山东大学 | Information hiding method, device and system based on void space pyramid |
| CN113254891B (en) * | 2021-05-17 | 2022-08-16 | 山东大学 | Information hiding method, device and system based on void space pyramid |
| CN113326531A (en) * | 2021-06-29 | 2021-08-31 | 湖南汇视威智能科技有限公司 | Robust efficient distributed face image steganography method |
| CN113538202A (en) * | 2021-08-05 | 2021-10-22 | 齐鲁工业大学 | Image steganography method and system based on generative steganography confrontation |
| CN114025053A (en) * | 2021-09-28 | 2022-02-08 | 上海电力大学 | Image steganography method combining high color similarity preprocessing |
| CN114025053B (en) * | 2021-09-28 | 2023-12-01 | 上海电力大学 | Image steganography method combined with high-color similarity preprocessing |
| CN114220443A (en) * | 2021-11-04 | 2022-03-22 | 合肥工业大学 | BN optimization SNGAN-based training method and system for adaptive audio steganography model |
| CN113989097B (en) * | 2021-12-30 | 2022-05-06 | 南京中孚信息技术有限公司 | Information steganography model training method, information steganography device and storage medium |
| CN113989097A (en) * | 2021-12-30 | 2022-01-28 | 南京中孚信息技术有限公司 | Information steganography model training method, information steganography device and storage medium |
| CN115022649A (en) * | 2022-05-17 | 2022-09-06 | 薇链信息技术有限公司 | Reversible neural network-based high-embedding-quantity and colorizable image steganography and recovery method |
| CN115022649B (en) * | 2022-05-17 | 2024-08-13 | 薇链信息技术有限公司 | High-embedding-capacity and colorable image steganography and restoration method based on reversible neural network |
| CN115086674A (en) * | 2022-06-16 | 2022-09-20 | 西安电子科技大学 | Image steganography method based on generation of countermeasure network |
| CN115086674B (en) * | 2022-06-16 | 2024-04-02 | 西安电子科技大学 | Image steganography method based on generation of countermeasure network |
| CN117292696A (en) * | 2023-10-08 | 2023-12-26 | 合肥工业大学 | End-to-end audio steganography method, system, storage medium and electronic equipment |
| CN117292696B (en) * | 2023-10-08 | 2024-03-12 | 合肥工业大学 | End-to-end audio steganography method, system, storage medium and electronic equipment |
| CN117132671A (en) * | 2023-10-27 | 2023-11-28 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Multi-task steganography method, system and medium based on depth self-adaptive steganography network |
| CN117132671B (en) * | 2023-10-27 | 2024-02-23 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Multi-task steganography method, system and medium based on depth self-adaptive steganography network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109587372B (en) | 2020-06-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109587372A (en) | A kind of invisible image latent writing art based on generation confrontation network | |
| Tolosana et al. | Deepfakes and beyond: A survey of face manipulation and fake detection | |
| Zhang | Deepfake generation and detection, a survey | |
| Lee et al. | Detecting handcrafted facial image manipulations and GAN-generated facial images using Shallow-FakeFaceNet | |
| Do et al. | Forensics face detection from GANs using convolutional neural network | |
| CN109800710A (en) | Pedestrian's weight identifying system and method | |
| CN111241291A (en) | Method and device for generating countermeasure sample by utilizing countermeasure generation network | |
| CN106599872A (en) | Method and equipment for verifying living face images | |
| CN111612799A (en) | Face data pair-oriented incomplete reticulate pattern face repairing method and system and storage medium | |
| Baek et al. | Generative adversarial ensemble learning for face forensics | |
| Asghar et al. | Edge–texture feature-based image forgery detection with cross-dataset evaluation | |
| CN112200075B (en) | Human face anti-counterfeiting method based on anomaly detection | |
| WO2022252372A1 (en) | Image processing method, apparatus and device, and computer-readable storage medium | |
| CN118196231B (en) | Lifelong learning draft method based on concept segmentation | |
| Zhao et al. | Detecting deepfake video by learning two-level features with two-stream convolutional neural network | |
| CN117437108A (en) | Watermark embedding method for image data | |
| Arora et al. | A review of techniques to detect the GAN-generated fake images | |
| Xiao et al. | Protecting the trust and credibility of data by tracking forgery trace based on GANs | |
| CN114596609A (en) | Audio-visual counterfeit detection method and device | |
| Khalid et al. | Deepfakes catcher: a novel fused truncated densenet model for deepfakes detection | |
| CN115936961B (en) | Steganalysis method, equipment and medium based on few-sample comparison learning network | |
| Yu et al. | Review of face presentation attack detection competitions | |
| CN111737688A (en) | Attack defense system based on user portrait | |
| Du et al. | WIP: Auditing Artist Style Pirate in Text-to-image Generation Models | |
| CN113642427A (en) | Image protection method for artificial intelligence counterfeiting |
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 |