CN112073732A - Method for embedding and decoding image secret characters of underwater robot - Google Patents
Method for embedding and decoding image secret characters of underwater robot Download PDFInfo
- Publication number
- CN112073732A CN112073732A CN202010871016.5A CN202010871016A CN112073732A CN 112073732 A CN112073732 A CN 112073732A CN 202010871016 A CN202010871016 A CN 202010871016A CN 112073732 A CN112073732 A CN 112073732A
- Authority
- CN
- China
- Prior art keywords
- loss
- image
- decoding
- embedding
- underwater robot
- 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.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/46—Embedding additional information in the video signal during the compression process
- H04N19/467—Embedding additional information in the video signal during the compression process characterised by the embedded information being invisible, e.g. watermarking
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/0021—Image watermarking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
- G06T9/002—Image coding using neural networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
- H04N19/91—Entropy coding, e.g. variable length coding [VLC] or arithmetic coding
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Signal Processing (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Editing Of Facsimile Originals (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a method for embedding and decoding image covert characters of an underwater robot, which comprises the following steps: combining four channels of an original image I and information M1 to be coded as input, and outputting a coded image O and a residual image R through a U-shaped network to obtain a first loss; inputting a coded image O, enhancing the robustness and generalization capability of the model, and calculating the cross entropy to obtain a second loss; weighting the first loss function and the second loss function, and analyzing and acquiring a function model with strong coding capability and low perception loss; training a BiSeNet semantic segmentation network, generating a large number of encoding graphs O, predicting candidate graph areas where the encoding graphs possibly appear, and then putting the candidate graph areas into a decoder for detection and judgment. By the method, the deep learning steganography is applied to the encoding, decoding and retrieval of the secret information of the bionic robot; performing deep learning steganography mode identification on the individual ID of global vision; and (4) encrypting and hiding the watermark of the consumption-level underwater robot.
Description
Technical Field
The invention relates to the technical field of image covert characters, in particular to a method for embedding and decoding image covert characters of an underwater robot.
Background
For three application scenes, namely bionic robot secret information carrying, global visual individual ID identification and consumer-grade underwater robot encrypted hidden watermark, efficient application of a deep neural network based on an autoencoder is one-time pioneering application, and no deep learning efficient solution with the same scene exists at present.
Disclosure of Invention
Aiming at the technical problems in the prior art, the invention provides a method for embedding and decoding stego characters in an image of an underwater robot, which can overcome the defects in the prior art.
In order to achieve the technical purpose, the technical scheme of the invention is realized as follows:
a method for embedding and decoding secret characters in an underwater robot image comprises the following steps:
s1: combining four channels of an original image I and information M1 to be coded as input, outputting a coding image O and a residual image R through a U-shaped network, and calculating I and O Watherstein losses to obtain a first loss;
s2: decoding the network, inputting a coded image O, enhancing the robustness and generalization capability of the model in an image data enhancement mode, outputting prediction information M2, M2 and M1 through a dense convolution network, comparing and performing supervised learning, and calculating cross entropy to obtain a second loss;
s3: weighting the first loss function and the second loss function, and analyzing and acquiring a function model with strong coding capability and low perception loss;
s4: training a BiSeNet semantic segmentation network, generating a large number of coding patterns O by using an ImageNet data set, embedding the coding patterns O into a sampling pattern of a DIV2K data set to be used as a synthetic data set for training, predicting an alternative pattern region P in which the coding patterns possibly appear, and then putting the alternative pattern region P into a decoder for detection and judgment.
Further, in step S2, the image data enhancement mode includes perspective transformation, motion blur, out-of-focus blur, color distortion, illumination deviation, gaussian noise, and JPEG encoding loss.
Further, in step S1, the loss of I and O watts is calculated as the first loss.
Further, in step S2, the cross entropy is taken as the second loss.
The invention has the beneficial effects that: by the method, the following steps are achieved: the deep learning steganography is applied to the encoding, decoding and retrieval of the secret information of the bionic robot; performing deep learning steganography mode identification on the individual ID of global vision; and (4) encrypting and hiding the watermark of the consumption-level underwater robot.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
Fig. 1 is a flowchart of a method for embedding and decoding stego characters in an underwater robot image according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments that can be derived by one of ordinary skill in the art from the embodiments given herein are intended to be within the scope of the present invention.
As shown in fig. 1, a method for embedding and decoding stego characters in an underwater robot image according to an embodiment of the present invention includes the following steps:
s1: combining four channels of an original image I and information M1 to be coded as input, outputting a coding image O and a residual image R through a U-shaped network, and calculating I and O Watherstein losses to obtain a first loss;
s2: decoding the network, inputting a coded image O, enhancing the robustness and generalization capability of the model in an image data enhancement mode, outputting prediction information M2, M2 and M1 through a dense convolution network, comparing and performing supervised learning, and calculating cross entropy to obtain a second loss;
s3: weighting the first loss function and the second loss function, and analyzing and acquiring a function model with strong coding capability and low perception loss;
s4: training a BiSeNet semantic segmentation network, generating a large number of coding patterns O by using an ImageNet data set, embedding the coding patterns O into a sampling pattern of a DIV2K data set to be used as a synthetic data set for training, predicting an alternative pattern region P in which the coding patterns possibly appear, and then putting the alternative pattern region P into a decoder for detection and judgment.
In an embodiment of the invention, in the step S2, the image data enhancement mode includes perspective transformation, motion blur, out-of-focus blur, color distortion, illumination deviation, gaussian noise, and JPEG encoding loss.
In a specific embodiment of the present invention, in step S1, the loss of I and O watts is calculated as the first loss.
In a specific embodiment of the present invention, in the step S2, the cross entropy is taken as the second loss.
The invention is further illustrated:
1. original image I and information M1 to be coded are merged into four channels as input, deep information is compressed and extracted through down sampling and up sampling of a U-shaped network, a coded graph O and a residual graph R are output, in order to reduce perception errors, Wastestan loss needs to be calculated through I and O, and the perception difference between the original graph and the coded graph is reduced.
2. Decoding the network, inputting a coded image O, enhancing the robustness and generalization capability of the model through image data enhancement modes such as perspective transformation, motion blur, defocus blur, color distortion, illumination deviation, Gaussian noise, JPEG coding loss and the like, finally outputting prediction information M2 through a dense convolution network, performing supervised learning through comparison with M1, and calculating cross entropy as a second loss.
3. And weighting the first two loss functions to obtain a model with strong coding capability and low perception loss.
4. Training a BiSeNet semantic segmentation network, generating a large number of code patterns O by using an ImageNet data set, embedding the code patterns O into a sampling pattern of a DIV2K data set to serve as a synthetic data set for training, and predicting an alternative pattern region P where the code patterns possibly appear, wherein the alternative pattern region is output in a multi-level IOU mode according to a cascade idea. And then put it into decoder detection judgment.
5. The whole network integrates encoding, decoding and detection, the encoder part is responsible for encoding information into a target image, the decoder is responsible for decoding the encoded image and recovering the information, and the detector is responsible for detecting the encoded image.
6. The whole model can encode 100-bit data, and 1080P images realize real-time detection and decoding.
In summary, with the above technical solution of the present invention, by the method, the following is achieved: the deep learning steganography is applied to the encoding, decoding and retrieval of the secret information of the bionic robot; performing deep learning steganography mode identification on the individual ID of global vision; and (4) encrypting and hiding the watermark of the consumption-level underwater robot.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (4)
1. A method for embedding and decoding secret characters in an underwater robot image is characterized by comprising the following steps:
s1: combining four channels of an original image I and information M1 to be coded as input, outputting a coding image O and a residual image R through a U-shaped network, and calculating I and O Watherstein losses to obtain a first loss;
s2: decoding the network, inputting a coded image O, enhancing the robustness and generalization capability of the model in an image data enhancement mode, outputting prediction information M2, M2 and M1 through a dense convolution network, comparing and performing supervised learning, and calculating cross entropy to obtain a second loss;
s3: weighting the first loss function and the second loss function, and analyzing and acquiring a function model with strong coding capability and low perception loss;
s4: training a BiSeNet semantic segmentation network, generating a large number of coding patterns O by using an ImageNet data set, embedding the coding patterns O into a sampling pattern of a DIV2K data set to be used as a synthetic data set for training, predicting an alternative pattern region P in which the coding patterns possibly appear, and then putting the alternative pattern region P into a decoder for detection and judgment.
2. The method for embedding and decoding stego characters in an underwater robot image as claimed in claim 1, wherein in the step S2, the image data enhancement mode includes perspective transformation, motion blur, out-of-focus blur, color distortion, illumination bias, gaussian noise and JPEG coding loss.
3. The method for embedding and decoding covert characters in underwater robot images as claimed in claim 1, wherein in said step S1, the loss of calculating I and O watts stanstein is taken as the first loss.
4. The method for embedding and decoding stego characters in underwater robot images as claimed in claim 1, wherein in step S2, cross entropy is taken as the second loss.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010871016.5A CN112073732A (en) | 2020-08-26 | 2020-08-26 | Method for embedding and decoding image secret characters of underwater robot |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010871016.5A CN112073732A (en) | 2020-08-26 | 2020-08-26 | Method for embedding and decoding image secret characters of underwater robot |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112073732A true CN112073732A (en) | 2020-12-11 |
Family
ID=73660048
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010871016.5A Pending CN112073732A (en) | 2020-08-26 | 2020-08-26 | Method for embedding and decoding image secret characters of underwater robot |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112073732A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114037596A (en) * | 2022-01-07 | 2022-02-11 | 湖南菠萝互娱网络信息有限公司 | End-to-end image steganography method capable of resisting physical transmission deformation |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110322435A (en) * | 2019-01-20 | 2019-10-11 | 北京工业大学 | A kind of gastric cancer pathological image cancerous region dividing method based on deep learning |
| CN110880193A (en) * | 2019-12-03 | 2020-03-13 | 山东浪潮人工智能研究院有限公司 | Image compression method using depth semantic segmentation technology |
| CN111028308A (en) * | 2019-11-19 | 2020-04-17 | 珠海涵辰科技有限公司 | Steganography and reading method for information in image |
| US20200184592A1 (en) * | 2017-05-19 | 2020-06-11 | Google Llc | Hiding Information and Images via Deep Learning |
-
2020
- 2020-08-26 CN CN202010871016.5A patent/CN112073732A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200184592A1 (en) * | 2017-05-19 | 2020-06-11 | Google Llc | Hiding Information and Images via Deep Learning |
| CN110322435A (en) * | 2019-01-20 | 2019-10-11 | 北京工业大学 | A kind of gastric cancer pathological image cancerous region dividing method based on deep learning |
| CN111028308A (en) * | 2019-11-19 | 2020-04-17 | 珠海涵辰科技有限公司 | Steganography and reading method for information in image |
| CN110880193A (en) * | 2019-12-03 | 2020-03-13 | 山东浪潮人工智能研究院有限公司 | Image compression method using depth semantic segmentation technology |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114037596A (en) * | 2022-01-07 | 2022-02-11 | 湖南菠萝互娱网络信息有限公司 | End-to-end image steganography method capable of resisting physical transmission deformation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Robust invisible video watermarking with attention | |
| KR100437187B1 (en) | Picture encoder, picture decoder, picture encoding method, picture decoding method, and medium | |
| Fernandez et al. | Watermarking images in self-supervised latent spaces | |
| CN111028308B (en) | Steganography and reading method for information in image | |
| Miller et al. | Dirty-paper trellis codes for watermarking | |
| Tang et al. | TSA-SCC: Text semantic-aware screen content coding with ultra low bitrate | |
| CN110009621B (en) | Tamper video detection method, tamper video detection device, tamper video detection equipment and readable storage medium | |
| CN111160313A (en) | Face representation attack detection method based on LBP-VAE anomaly detection model | |
| Eid et al. | Digital image steganalysis: current methodologies and future challenges | |
| CN116091288A (en) | Diffusion model-based image steganography method | |
| Makar et al. | Interframe coding of canonical patches for mobile augmented reality | |
| CN112073732A (en) | Method for embedding and decoding image secret characters of underwater robot | |
| Ranjbar Alvar et al. | Joint image compression and denoising via latent-space scalability | |
| Maity et al. | Genetic algorithms for optimality of data hiding in digital images | |
| Dai et al. | HEVC Video Steganalysis Based on PU Maps and Multi-Scale Convolutional Residual Network | |
| US6263115B1 (en) | Method and apparatus for encoding a binary shape signal | |
| CN117079354A (en) | Deep forgery detection classification and positioning method based on noise inconsistency | |
| WO2023203509A1 (en) | Image data compression method and device using segmentation and classification | |
| Pomponiu et al. | Data hiding in the wild: where computational intelligence meets digital forensics | |
| CN114418821A (en) | Blind watermark processing method based on image frequency domain | |
| CN115699110A (en) | Segmentation mask generation in alpha channel based on automatic encoder | |
| Cui et al. | Deeply‐Recursive Attention Network for video steganography | |
| Singh et al. | Data hiding in videos using background subtraction | |
| Ben Jabra et al. | Deep Learning-Based Watermarking Techniques Challenges: A Review of Current and Future Trends | |
| Chang et al. | Reversible data hiding in the VQ-compressed domain |
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 |