CN109543425A - A kind of Image Data Hiding Methods based on tensor resolution - Google Patents
A kind of Image Data Hiding Methods based on tensor resolution Download PDFInfo
- Publication number
- CN109543425A CN109543425A CN201811309996.9A CN201811309996A CN109543425A CN 109543425 A CN109543425 A CN 109543425A CN 201811309996 A CN201811309996 A CN 201811309996A CN 109543425 A CN109543425 A CN 109543425A
- Authority
- CN
- China
- Prior art keywords
- image
- tensor
- singular
- embedding information
- mode
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/602—Providing cryptographic facilities or services
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Bioethics (AREA)
- Image Processing (AREA)
- Editing Of Facsimile Originals (AREA)
Abstract
The invention discloses a kind of Image Data Hiding Methods based on tensor resolution, Higher-order Singular value decomposition is applied to the image of required embedding information, the corresponding left singular matrix element of minimum singular value that required embedding information replacement tensor resolution is obtained, reconstructs original image by replaced singular matrix.This method comprises: the image of required embedding data is put into a three rank square tensors;Square tensor is decomposed using Higher-order Singular value decomposition method, obtains singular value and singular matrix of the core tensor sum in each mode;Required embedding information is replaced into the corresponding left singular matrix element of minimum singular value under a certain mode obtained by tensor resolution;By three left singular matrixs and core tensor three rank tensors of reconstruct after embedding information;The image after embedding information is extracted in three rank tensors.The present invention hides tensor resolution applied to image data, improves the safety of information transmission, while not will cause the decline of picture quality after extracting embedding information.
Description
Technical field
The present invention relates to data hiding technique more particularly to a kind of Image Data Hiding Methods based on tensor resolution.
Background technique
Data hiding technique is a kind of comprehensive technology, it be related to anthropophysiology, computer graphics, cryptography,
The knowledge of multiple subjects such as information processing, what it was mainly studied is how confidential information to be hidden in public information, is then led to
Public information is crossed to transmit confidential information.General data hiding technique can change initial data during information is embedded in,
It will cause the damage of initial data after extracting embedding information, therefore lossless data hiding technique becomes data hiding technique
Focus on research direction.
Image data concealing technology is a branch of data hiding technique, and confidential information is embedded in a certain disclosed by it
In image, confidential information is transmitted by the image, this transmission mode will not cause attacker's note that relative to password skill
Art is safer.Existing Image Data Hiding Methods:
Fridrich.J proposes a kind of reversible information hidden method based on lossless data compression.Lossless data compression side
Method is mainly to extract the LSB information of original image and compressed, and the space generated after compression is used to be embedded in hiding information, the party
Method does not destroy raw information, and the space obtained by compression raw information is come embedding information, but the sky generated in compression process
Between it is smaller, data insertion rate is lower, and poor robustness.
Tian proposes a kind of reversible information hidden method based on difference expansion.The basic thought of difference expansion is to pass through
Difference between the pixel of extending neighboring carrys out hiding information.This method that Tian is proposed is by calculating extension difference k 'l=
2kl+wlCarry out embedding data, but this method is easy to cause pixel value to overflow, and can solve the problem of pixel is overflowed using positioning figure
Realize that image restores, but it is larger to embedding information capacity impact.
Ni proposes a kind of reversible information hidden method based on Histogram adjustment.The basic thought of Histogram adjustment is logical
It crosses and the histogram of original image is adjusted, using zero point redundancy come embedding information.This method that Ni is proposed utilizes image
Three maximum points and smallest point carry out the insertion of information in histogram, and the information capacity of this method insertion is big, but complexity compared with
Height, and be not the best position of embedding information at maximum point.
The present invention provides a kind of Image Data Hiding Methods based on tensor resolution, and original RGB image is put into one
In three rank square tensors, position is put into as key, the tensor is decomposed to obtain three using Higher-order Singular value decomposition
Singular matrix under a mode is unusual using a left side corresponding to the minimum singular value in the mode determined by original image size
The information of required insertion is replaced the column element by vector, by decomposing the left surprise under obtained core tensor and three mode
Different matrix (left singular matrix including embedding information in the mode) reconstructs three rank tensors, obtains embedding information by location key
RGB image afterwards, doing Higher-order Singular value decomposition again to the RGB image can be obtained embedding information (minimum singular value in the mode
Corresponding left singular matrix element), this method influence caused by picture quality after extracting embedding information is minimum.
Summary of the invention
The invention aims to make full use of the space structure in RGB image, attack can be resisted to a certain extent,
Increase embedding information safety, while to embedding information extract after, picture quality is held essentially constant.
To achieve the goals above, the present invention provides a kind of Image Data Hiding Methods based on tensor resolution, the party
Method includes:
Original RGB image is put into the three rank square tensors an of random value, the position being put into is as one
Key;
Three rank tensors are decomposed using Higher-order Singular value decomposition, obtain a core tensor sum respectively in three mode
Under singular matrix (including three left singular matrixs, three singular value matrixs, three right singular matrixs);
Singular value matrix after decomposition obtains under selection mode determined by original image size finds wherein the smallest
The information of required insertion is replaced the column element and obtains left surprise new under the mode by left singular matrix element corresponding to singular value
Different matrix;
Using the left singular matrix and core tensor reconstruct Hui Sanjie tensor under three mode, passes through cipher key-extraction and be embedded in letter
Image after breath;
Singular matrix is obtained by Higher-order Singular value decomposition to the image of embedding information, the minimum found under the mode is unusual
The corresponding left singular matrix element of value, which is embedding information;
After extracting information, tensor is reconstructed using the left singular matrix decomposed under obtained core tensor and three mode, is passed through
Location key finds the image after extracting information.
Wherein, according to the three ranks square tensor, comprising:
Three rank square tensor sizes are determined according to selected image: a=max (size (I)), I are selected image;
Tensor element value is in 0-255;
Image is put into the position of three rank square tensors as secret key;
Wherein, three rank square tensors are decomposed using higher order singular value according to described, comprising:
Three rank square tensors are decomposed to obtain the expansion matrix under three mode of a core tensor sum by Tucker;
SVD is to the expansion matrix under three mode to decompose to obtain singular matrix (left singular matrix, singular value matrix, the right side
Singular matrix), the corresponding matrix size of three mode is identical.
Wherein, image is extracted according to the embedding information and three rank tensors of reconstruct, comprising:
It finds left singular matrix column vector corresponding to minimum singular value, the information that will be embedded in the mode and replaces the column
Vector;
By the left singular matrix and core tensor three rank tensors of reconstruct under the left singular matrix and other mode;
Original image after extracting embedding information by location key.
Wherein, according to the extraction embedding information, comprising:
Embedding information image is put into original three ranks square tensor by location key;
Higher-order Singular value decomposition is done to the three ranks tensor;
The corresponding left singular matrix element of minimum singular value under the mode is found, which is embedding information;
By under the left singular matrix and other mode after extraction information left singular matrix and core tensor reconstruct three ranks
Amount;
The image after extracting embedding information is found in three rank square tensors by location key.
Based on foregoing invention method, there is difference and contribution outstanding to be compared with the conventional method:
1, the present invention relates to images is stated with tensor form, makes full use of the space structure of RGB image, the tensor formed
It decomposes to obtain three mode by Tucker, it, can be next by being unfolded under the mode determined by original image size
The SVD of step retains more image details in decomposing, and the influence caused by picture quality is minimum after embedding information is extracted.
2, three ranks square tensor of the present invention, the matrix after three obtained modal expanding is decomposed by Tucker
Size is identical, and when under attack, which mode attacker, which cannot intuitively find out, is protected determined by original image size
The information being embedded in singular matrix under the mode has been protected, the safety of the information in transmission process is improved.
Detailed description of the invention:
Technical solution of the present invention is described further below in conjunction with attached drawing.
Fig. 1 is the flow chart of the Image Data Hiding Methods based on tensor resolution of the embodiment of the present application.
Fig. 2 is the flow chart that the embodiment of the present application image is put into three rank square tensors.
Fig. 3 is that the information of the embodiment of the present application is embedded in flow chart.
Fig. 4 is the information extraction flow chart of the embodiment of the present application.
Specific embodiment:
The object, technical solutions and advantages of the present invention will be discussed in detail by specific embodiment and attached drawing below.
Fig. 1 show the whole flow process schematic diagram of the embodiment of the present application, is embodied as follows:
Step S110, original RGB image are put into the three rank square tensors that one is established at random.
Step S120 does Higher-order Singular value decomposition to three rank tensors and obtains singular values and singular vectors.
Required embedding information is replaced the member of left singular matrix corresponding to minimum singular value under a certain mode by step S130
Element.
The left singular matrix replaced and core tensor are reconstructed tensor, image are found in tensor by step S140.
Step S150, the image after obtaining original three ranks square tensor sum embedding information, does Higher-order Singular value decomposition i.e.
Embedding information can be obtained.
Several committed steps of the embodiment of the present application are described in detail below.
One, Fig. 2 indicates the foundation of three rank square tensors
The embodiment of the present application is by original RGB image IOIt is put into the three rank square tensor T an of random value,
In:
Original RGB image IOSize is n × m × 3;
Three rank square tensor T of random valueO;
Three rank square tensor TOIf size be n > m, tensor size is n × n × n, and corresponding mode is the first mould
State, if n < m, for m × m × m, corresponding mode is second mode (present embodiment assumes that n > m, chooses first mode);
The element of filling is between 0-255;;
In three rank tensor TOIn select some position at random and be put into original RGB image IOIt obtains being put into three ranks after image
Measure TE;
The position is a key, is determined by encipherer.
Two, Fig. 3 indicates the process of hiding information insertion
The embodiment of the present application is using minimum singular value come embedding information, in which:
Step S310, using Higher-order Singular value decomposition to three rank tensor TEIt is decomposed;
Tucker is to tensor to decompose to obtain the matrix under each mode of core tensor sum;
Tucker decomposable process may be expressed as: TE=[S, A, B, C], wherein S is core tensor, and A, B, C is respectively three
Matrix under mode, it is approximately equal to TE≈S×1AU×2BU×3CU.SVD decomposition is done to A, B, C respectively, is obtained under each mode
Singular matrix:
[AU, AS, AV]=SVD (A);
[BU, BS, BV]=SVD (B);
[CU, CS, CV]=SVD (C);
Wherein subscript U, S, V respectively indicate the left singular matrix under the mode, singular value matrix, right singular matrix;
Step S320 is chosen at the singular value matrix A that mode is once decomposed by SVDS, pick out therein minimum odd
Different value σmin, in left singular matrix AUIn find out by σminCorresponding column vector;
Step S330, the information that will be embedded in replace the column vector, the left singular matrix A ' after obtaining information insertionU;
Step S340, the left singular matrix A ' by core tensor S, after one embedding information of modeUAnd mode two and mode
Left singular matrix B under threeU, CUThree rank tensors after obtaining information insertion;
Image I using location key, after extracting information insertion in three rank tensors of reconstructR。
Three, Fig. 4 indicates the process that embedding information is extracted
Step S410, using location key, by the image I of embedding informationRIt is put into three initial rank square tensor TOIt obtains
Three new rank tensor TN;
Step S420, to TNDo Higher-order Singular value decomposition (with to TEThe decomposition done is identical), obtain a core tensor S2With
Singular matrix under each mode includes three mode, nine singular matrix;
Step S430 finds the left singular matrix of mode once, is exactly by the corresponding column singular vector of minimum singular value
The information of insertion;
Step S440, after extracting embedding information, by core tensor S2, the unusual square in a left side for being extracted information in mode one
Corresponding left singular matrix reconstructs three rank tensor T under battle array and other two modeK;
Step S450, using location key, three rank tensor T after reconstitutionKMiddle extraction image obtains being extracted insertion letter
The image of breath, the picture quality are similar to original image quality.
Claims (5)
1. a kind of Image Data Hiding Methods based on tensor resolution, for passing through image transmitting information, which is characterized in that described
Method the following steps are included:
Three rank square tensors of the element between (0-255) are established, original RGB image is put into the three ranks tensor
In;
Three rank tensors are decomposed using Higher-order Singular value decomposition, obtain the unusual square under three mode of a core tensor sum
Battle array (including three left singular matrixs, three singular value matrixs, three right singular matrixs);
The singular value matrix decomposed under first mode is selected to find the unusual square in a left side corresponding to wherein the smallest singular value
Array element element, replaces the column element for the information of required insertion and obtains left singular matrix new under the mode;
Information is extracted using the left singular matrix and core tensor reconstruct Hui Sanjie tensor under three mode, then from three rank tensors
Image after insertion;
Image after embedding information is put into original three ranks tensor through the available embedding information of Higher-order Singular value decomposition.
2. the Image Data Hiding Methods according to claim 1 based on tensor resolution, which is characterized in that original image is put
Enter to three rank square tensors, comprising:
The RGB image of required embedding information is put into three rank square tensors;
The position being put into is as a key;
Three rank square tensor element values are between 0-255;
Size is the maximum value of picture size.
3. the Image Data Hiding Methods according to claim 1 based on tensor resolution, which is characterized in that odd using high-order
Three rank tensors are decomposed in different value decomposition, comprising:
Three rank tensors after being put into image are decomposed using Higher-order Singular value decomposition;
Obtain its singular matrix (including singular value matrix, left singular matrix, right surprise in three mode of a core tensor sum
Different matrix);
Choose left singular matrix element corresponding to the minimum singular value that the mode decomposition as corresponding to original image size obtains;
Required embedding information is replaced into the column element, obtains the left singular matrix under the mode after embedding information.
4. the Image Data Hiding Methods according to claim 1 based on tensor resolution, which is characterized in that after replacement
Left singular matrix and kernel matrix reconstructed image, comprising:
Utilize the left singular matrix and core under the left singular matrix and other mode under the mode after embedding information
Amount three rank square tensors of reconstruct;
RGB image after finding out embedding information with location key in three rank tensors of reconstruct;
RGB image size after embedding information is consistent with original RGB image.
5. the Image Data Hiding Methods according to claim 1 based on tensor resolution, which is characterized in that embedding information
Extraction is done Higher-order Singular value decomposition by the three rank tensors for being put into the image of embedding information and is obtained, comprising:
The image after embedding information is put into initial three ranks tensor using location key;
Higher-order Singular value decomposition is done to three rank tensors after image are put into;
Embedding information is the corresponding singular vector of minimum singular value in the mode as corresponding to original image size.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811309996.9A CN109543425B (en) | 2018-10-26 | 2018-10-26 | Image data hiding method based on tensor decomposition |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811309996.9A CN109543425B (en) | 2018-10-26 | 2018-10-26 | Image data hiding method based on tensor decomposition |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109543425A true CN109543425A (en) | 2019-03-29 |
CN109543425B CN109543425B (en) | 2023-03-10 |
Family
ID=65846114
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811309996.9A Active CN109543425B (en) | 2018-10-26 | 2018-10-26 | Image data hiding method based on tensor decomposition |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109543425B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112311953A (en) * | 2020-09-28 | 2021-02-02 | 北京临近空间飞行器系统工程研究所 | Reversible image information hiding method and device, electronic equipment and medium |
CN116912107A (en) * | 2023-06-13 | 2023-10-20 | 重庆市荣冠科技有限公司 | DCT-based weighted adaptive tensor data completion method |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100183190A1 (en) * | 2007-09-21 | 2010-07-22 | Kazuo Ohzeki | Electronic watermark embedding apparatus and method and electronic watermark detecting apparatus and method |
WO2011026365A1 (en) * | 2009-09-03 | 2011-03-10 | 中兴通讯股份有限公司 | Method and system for embedding and extracting image digital watermark |
WO2014071714A1 (en) * | 2012-11-06 | 2014-05-15 | 中国科学院自动化研究所 | Watermark embedding and extracting method |
AU2017100244A4 (en) * | 2017-02-13 | 2017-03-30 | Macau University Of Science And Technology | Methods and Apparatus for Color Image Watermarking |
CN106791872A (en) * | 2016-11-18 | 2017-05-31 | 南京邮电大学 | Information concealing method based on SVD |
CN107247688A (en) * | 2017-06-02 | 2017-10-13 | 上海德衡数据科技有限公司 | A kind of multichannel data compressed sensing cube framework |
CN107274333A (en) * | 2017-06-21 | 2017-10-20 | 鲁东大学 | A kind of spatial domain coloured image blind watermark method for merging singular value decomposition |
CN107292806A (en) * | 2017-06-28 | 2017-10-24 | 南京师范大学 | A kind of remote sensing image digital watermark embedding and extracting method based on quaternion wavelet |
CN107835427A (en) * | 2017-11-29 | 2018-03-23 | 福建北卡科技有限公司 | A kind of color image information hidden method based on contourlet and SVD |
CN108681982A (en) * | 2018-04-04 | 2018-10-19 | 微梦创科网络科技(中国)有限公司 | A kind of embedding algorithm and extracting method and system based on singular value decomposition |
-
2018
- 2018-10-26 CN CN201811309996.9A patent/CN109543425B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100183190A1 (en) * | 2007-09-21 | 2010-07-22 | Kazuo Ohzeki | Electronic watermark embedding apparatus and method and electronic watermark detecting apparatus and method |
WO2011026365A1 (en) * | 2009-09-03 | 2011-03-10 | 中兴通讯股份有限公司 | Method and system for embedding and extracting image digital watermark |
WO2014071714A1 (en) * | 2012-11-06 | 2014-05-15 | 中国科学院自动化研究所 | Watermark embedding and extracting method |
CN106791872A (en) * | 2016-11-18 | 2017-05-31 | 南京邮电大学 | Information concealing method based on SVD |
AU2017100244A4 (en) * | 2017-02-13 | 2017-03-30 | Macau University Of Science And Technology | Methods and Apparatus for Color Image Watermarking |
CN107247688A (en) * | 2017-06-02 | 2017-10-13 | 上海德衡数据科技有限公司 | A kind of multichannel data compressed sensing cube framework |
CN107274333A (en) * | 2017-06-21 | 2017-10-20 | 鲁东大学 | A kind of spatial domain coloured image blind watermark method for merging singular value decomposition |
CN107292806A (en) * | 2017-06-28 | 2017-10-24 | 南京师范大学 | A kind of remote sensing image digital watermark embedding and extracting method based on quaternion wavelet |
CN107835427A (en) * | 2017-11-29 | 2018-03-23 | 福建北卡科技有限公司 | A kind of color image information hidden method based on contourlet and SVD |
CN108681982A (en) * | 2018-04-04 | 2018-10-19 | 微梦创科网络科技(中国)有限公司 | A kind of embedding algorithm and extracting method and system based on singular value decomposition |
Non-Patent Citations (3)
Title |
---|
CLIFFORD BERGMAN ET AL.: "Unitary Embedding for Data Hiding with the SVD", 《SECURITY,STRGANOGRAPHY,AND WATERMARKING OF MULTIMEDIA CONTENTS VII》 * |
严明等: "基于分块DCT和Tucker分解的彩色图像数字水印", 《计算机应用》 * |
夏建平等: "基于小波变换和Tucker分解的彩色图像数字水印", 《计算机应用研究》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112311953A (en) * | 2020-09-28 | 2021-02-02 | 北京临近空间飞行器系统工程研究所 | Reversible image information hiding method and device, electronic equipment and medium |
CN116912107A (en) * | 2023-06-13 | 2023-10-20 | 重庆市荣冠科技有限公司 | DCT-based weighted adaptive tensor data completion method |
CN116912107B (en) * | 2023-06-13 | 2024-04-16 | 万基泰科工集团数字城市科技有限公司 | DCT-based weighted adaptive tensor data completion method |
Also Published As
Publication number | Publication date |
---|---|
CN109543425B (en) | 2023-03-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Lin et al. | Invertible secret image sharing with steganography | |
Thien et al. | An image-sharing method with user-friendly shadow images | |
Yang et al. | Reversible steganography based on side match and hit pattern for VQ-compressed images | |
CN106791872B (en) | Information concealing method based on SVD | |
CN107689026B (en) | Reversible steganography method based on optimal coding | |
CN103426141B (en) | A kind of image content authentication method and system | |
Chang et al. | A sudoku-based secret image sharing scheme with reversibility | |
CN103501400B (en) | Based on the reversible information hidden method of parabola interpolation image | |
CN103414840A (en) | Data hiding method based on image set | |
Goli et al. | Introducing a new method robust against crop attack in digital image watermarking using two-step sudoku | |
CN109543425A (en) | A kind of Image Data Hiding Methods based on tensor resolution | |
Ou et al. | Reversible AMBTC-based secret sharing scheme with abilities of two decryptions | |
Chen et al. | High-capacity reversible data hiding in encrypted image based on Huffman coding and differences of high nibbles of pixels | |
CN113115053A (en) | Image encryption method based on integer wavelet transform and compressed sensing | |
Yang et al. | Tampering detection and content recovery for digital images using halftone mechanism | |
CN108521535B (en) | A kind of Information hiding transmission method based on image blend processing | |
Xiao et al. | High-payload completely reversible data hiding in encrypted images by an interpolation technique | |
CN110766597B (en) | Picture camouflage method and device, electronic equipment and storage medium | |
CN115330583A (en) | Watermark model training method and device based on CMYK image | |
CN108876721A (en) | Super-resolution image reconstruction method and system based on course learning | |
Li et al. | High‐Capacity Reversible Data Hiding in Encrypted Images by Information Preprocessing | |
Chiu et al. | An XOR-based progressive visual cryptography with meaningful shares | |
Wang et al. | A novel (n, n) secret image sharing scheme based on Sudoku | |
CN105046633B (en) | A kind of image lossless transform method | |
CN110706145B (en) | Reversible remote sensing image information hiding processing method and device |
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 |