CN110428355A - A kind of Watermarking for Color Images method based on compressed sensing and NSCT - Google Patents
A kind of Watermarking for Color Images method based on compressed sensing and NSCT Download PDFInfo
- Publication number
- CN110428355A CN110428355A CN201910608251.0A CN201910608251A CN110428355A CN 110428355 A CN110428355 A CN 110428355A CN 201910608251 A CN201910608251 A CN 201910608251A CN 110428355 A CN110428355 A CN 110428355A
- Authority
- CN
- China
- Prior art keywords
- watermark
- block
- embedded
- color
- watermarking
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- 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
- G06T1/005—Robust watermarking, e.g. average attack or collusion attack resistant
-
- 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
- G06T1/005—Robust watermarking, e.g. average attack or collusion attack resistant
- G06T1/0057—Compression invariant watermarking
Abstract
A kind of Watermarking for Color Images method based on compressed sensing and NSCT.This method decomposes color host image R, G, B, to all three color component homalographic non-overlap piecemeals, m piecemeal is randomly choosed in all piecemeals and constitutes insertion array Q1.NSCT decomposition is carried out to embedded blocks all in Q1, its low frequency component LU is taken to decompose, diagonal matrix SVD is decomposed and extracts singular value；Watermarking images homalographic is divided into m/2 block, with same every piece of compression sampling operator extraction of measured value.The measured value of jth block watermark is respectively embedded into array Q1 in the singular value of jth block and 2j block, as long as rule selection embedded block when being embedded according to watermark, extracts watermark information.The present invention ensure that watermark large capacity is embedded in, while compressed sensing increases the safety of watermark again by reducing data redudancy to the perception of watermarking images splits' positions.This method is effective against the geometric attacks such as shearing, scaling, JPEG compression, high-pass filtering, salt-pepper noise, suitable for the copyright protection to color digital image.
Description
Technical field
The invention belongs to field of information security technology, are related to a kind of color image number water based on compressed sensing and NSCT
Impression method.
Background technique
The digitlization of information is an important feature of current social development, with digital information (image, audio, video,
Digital publishing etc.) it is in explosive increase.People are more and more convenient by internet acquisition digital information, while pirate cost
It is more and more cheaper.The digital watermarking effective means anti-fake as intellectual property protection and digital medium information, precisely in order to dimension
It protects the equity of original people and copyright buyer and is born.
Since what color image included contains much information, it is increasingly becoming the hot spot of the research of watermark in recent years.And existing colour
Arithmetic on Digital Watermarking of Image is frequently present of the characteristics of small watermark capacity, insertion point concentration, when making its anti-shearing equal geometric attack
Ability dies down, even as extract less than watermark the problems such as.
Summary of the invention
The object of the present invention is to according to existing Color Images Watermarking Algorithm there are the problem of, in the R of color image,
G, watermark embedded block array is randomly choosed on B component, and watermarking images are embedded in array after compressed sensing extracts measured value repeatedly
In block.One aspect of the present invention solves the problems in watermark insertion excessively collection, on the other hand reduces watermark by compressed sensing
The redundancy of information increases the information content of watermark insertion, and compressed sensing also further increases the safety of watermark information
(compressed sensing observing matrix seed and block size all can serve as key, person with foresight's condition as watermark extracting).
Technical solution of the present invention: Watermarking for Color Images new method of the invention based on compressed sensing and NSCT.Including
Digital watermark embedding and digital watermark extracting.The present invention selects color host image to carry out RGB decomposition, non-to each color homalographic
Overlap partition, randomly choosed from all piecemeals m (m be natural number and be 2 integral multiple) a embedded block constitutes insertion array, right
Every piece of progress NSCT transformation in array carries out its low frequency component and obtains diagonal matrix after LU is decomposed, then to diagonal matrix SVD points
Solution obtains singular value.Gray-level Watermarking image homalographic non-overlap is divided into m/2 block, measured value is obtained to every piece of compressed sensing,
Measured value is embedded into the singular value of insertion array according to established rule, because the embedded block number of insertion array is watermark block
2 times, so ensure that each watermark block is embedded into 2 times；In watermark detection, as long as to being carried containing watermark in insertion array
Body block carries out watermark extracting, can obtain complete watermark information.In this programme be embedded in array by three color component piecemeals with
Machine, which is chosen, to be constituted, and ensure that discrete be distributed in entire carrier image of the watermark block of insertion can be reasonably resistant to all kinds of geometry and attack
It hits.
Steps are as follows for data waterprint embedded method in the present invention:
(1) color host image O is carried out R, G, B to decompose, to each color component non-overlap homalographic piecemeal, obtains image
Set of blocks Q, randomly choosed from set Q m (m be natural number and be 2 integral multiple) a piecemeal constitutes insertion array Q1, by this m
The home position of a data block is saved as key key1.
(2) to array Q1In every piece carry out 3 grades of NSCT and decompose, it is assumed that use XiRepresent insertion array Q1In i-th (i=1,
The low frequency component matrix of 2 ... m) blocks, carries out LU decomposition to it:
Wherein l21、l31、l32For the non-zero numerical value of lower triangular matrix L, d1、d2、d3For the non-zero numerical value of diagonal matrix D, u12、
u13、u23For the non-zero value of upper triangular matrix U.
Diagonal matrix D can be obtained by formula (1)i, the number that i is every piece, to matrix DiSVD is carried out to decompose:
Wherein T is conjugate transposition, UiAnd Vi TFor orthogonal matrix, SiFor diagonal matrix, λi1,λi2…λinFor matrix SiIt is unusual
Value.Save λi1,λi2,…,λinValue be key2, in case extract watermark when use.Save Ui, Vi TValue, in case watermark insertion when
It uses.
(3) to insertion array Q1In all piecemeals low frequency sub-band matrix execute a step (2).
(4) non-overlap piecemeal is carried out to Gray-level Watermarking image W, block count m/2 is array Q1The half of middle embedded block number.
(5) the identical compression sampling operator Φ of each watermark block is observed, corresponding measured value, Φ and watermarking images can be obtained
Block size is saved as key key3.The population of measured values of each piecemeal is n, uses Yj(j=1,2...m/2) jth is indicated
The measured value of a watermark piecemeal.
(6) measured value of watermark jth (j=1,2 ..., m/2) block is respectively embedded in array Q1Middle jth block and 2j block
The diagonal matrix S of carrier componentiSingular value in.
α in formulajFor jth block and the embedment strength of 2j block, α is savedjValue be key4, in case watermark extracting use.
WithInstead of the S in formula (2)i
(7) it usesInstead of the D in formula (1)i
(8) it usesInstead of array Q1In every piece of embedded block low frequency component, it is inverse together with its other original frequency domain components
Block containing watermark embedded data newly is obtained after NSCT transformation.
(9) according to key key1, by each containing watermark according to block, its data block in former R, G, B component is replaced respectively,
New R containing watermark, G, B component are formed, the O of carrier image containing watermark is synthesizedW。
The step of digital watermarking extracting method, is as follows in the present invention:
(1) by aqueous printed color picture OwIt carries out R, G, B to decompose, size when respectively to R, G, B component by watermark insertion
Homalographic non-overlap piecemeal chooses piecemeal according to key key1 and constitutes insertion array Qw。
(2) by array QwIn every piece carry out 3 grades of NSCT and decompose, it is assumed that use XWiRepresent insertion array QwIn i-th (i=1,
The low frequency component matrix of 2 ... m) blocks, carries out LU decomposition to it:
(3) diagonal matrix D can be obtained by formula (7)Wi, the number that i is every piece, to array QwIn every block diagonal matrix DWiIt carries out
SVD is decomposed:
Wherein λWi1,λWi2…λWinThe singular value decomposed for SVD
(4) D before the watermark insertion known to key key2iCarry out the singular value of SVD decomposition.IfFor j-th of watermark block
Measured value, extract
αjValue can be obtained by key4.
(5) compressed sensing reconstructuring water-mark image W* is carried out to extracted all m/2 measured values.
The beneficial effects of the present invention are: the present invention passes through the digital watermark embedding and digital watermark extracting energy to color image
Complete watermark information is obtained, this method does not need to provide initial carrier image, will contain watermarking images when carrying out watermark detection
Three color components by insertion when with size homalographic divide, according to key key1 choose embedded block, to each piece of progress watermark
The inverse transformation of insertion obtains watermarking images.The present invention is protected by reducing data redudancy to the perception of watermarking images splits' positions
The insertion of watermark large capacity is demonstrate,proved.Compressed sensing increases the safety of watermark again simultaneously.The method of the present invention be effective against shearing,
The geometric attacks such as scaling, brightness adjustment, lossless compression, high-pass filtering, suitable for the copyright protection to color digital image.
Detailed description of the invention
Fig. 1 is flow chart of the invention,
Fig. 2 (a) carrier image Kid Fig. 2 (b) carrier image Lena,
Fig. 3 watermarking images,
Fig. 4 (a) Kid containing watermarking images Fig. 4 (b) Lena containing watermarking images.
Specific embodiment
The present invention is as follows by case verification:
Color host image is Kid, Lena of 512 × 512 pixels, and after R, G, B are decomposed, each color component is with size for 64
The segmentation of × 64 size non-overlaps randomly chooses 64 pieces of compositions from all piecemeals and is embedded in array Q1.Watermarking images are 128 × 128
Grey cameraman image, homalographic non-overlap are divided into 32 pieces.In order to guarantee the safety of watermark, each watermark block is embedding
Enter twice, so the number of embedded block is 2 times of watermark block.
In terms of evaluating watermark extracting effect, similarity judge, NC are carried out using normalizated correlation coefficient NC in experiment
The bigger similarity of watermark and original watermark for illustrating to extract of value it is higher.
Wherein, W expression original watermark, the watermark that W ' expression is extracted, M and N are respectively the length and width pixel value of watermarking images.
The NC value that watermark is extracted after different type is attacked is as follows:
Attack type: NC (being Kid and Lena value respectively)
Claims (3)
1. a kind of be embedded in extracting method based on the Watermarking for Color Images of compressed sensing and wavelet transformation, which is characterized in that institute
The method of stating includes digital watermark embedding and digital watermark extracting；The method selects gray level image as watermark, by watermarking images
Homalographic is divided into p block, after extracting high fdrequency component to every piece of wavelet decomposition, with same every piece of compression sampling operator extraction of measurement
Value；Color host image carries out R, G, B color decomposition, is divided into m block (m=3p) to each color component homalographic non-overlap,
It respectively selects p piecemeal to constitute insertion array Q1 from each color piecemeal respectively according to rule, embedded blocks all in Q1 is carried out
Wavelet transformation chooses compress high frequency components perception and extracts measured value, the measured value of watermark block is respectively embedded into according to certain rule
In the measured value that each piece of array Q1, in watermark extracting, as long as rule selection embedded block when being embedded according to watermark, extracts watermark
Information.
2. a kind of be embedded in based on the Watermarking for Color Images of compressed sensing and wavelet transformation according to claim 1 is extracted
Method, which is characterized in that steps are as follows for the data waterprint embedded method:
(1) non-overlap piecemeal, block count p (p is positive integer) are carried out to Gray-level Watermarking image W；
(2) to all watermark piecemeal Wi(i=1,2 ... p) do wavelet transformation, take high frequency coefficient matrix Hi；
(3) by sparse basis Ψ and observing matrix Φ, compressed sensing transformation is done to the high frequency coefficient matrix of all watermark piecemeals, is obtained
The measured value of each piecemeal, Ψ and Φ are saved as key key1；
Wherein (i=1,2 ... p), and n is the number of measurements of watermark block；
(4) color host image O is carried out R, G, B to decompose, m (m=3p) block is divided into each color component non-overlap homalographic,
Each piecemeal number Akj(k=(R, G, B), j=1,2 ... m), respectively selects p piecemeal group from tri- color piecemeals of R, G, B respectively
At 3 insertion arraysThe selection principle of array: the number A of selected all piecemealskj,
J value does not allow identical)；It is saved the home position of this m data block as key key2；
(5) to insertion arrayIn every piece do wavelet transformation, take every piece of high frequency coefficient matrix pass through sparse basis ΨkAnd observing matrix
ΦkCompressed sensing transformation is done, the measured value of each piecemeal, Ψ are obtainedkAnd ΦkIt is saved as key key3；
Wherein, k=(R, G, B), (j=1,2 ..., m), n are the number of measured value obtained by each matrix；Save all Ykj's
Value, in case being used when watermark extracting；
(6) measured value of watermark i-th (i=1,2 ..., p) block is respectively embedded in arrayI-th piece in；
K=(R, G, B), αiFor the embedment strength of the i-th watermark block, α is savediValue be key4, in case watermark extracting use；
(7) measured value after insertion watermark is subjected to compressed sensing reconstruct, the high-frequency sub-band coefficient matrix after obtaining insertion watermark
H′k；
(8) to high-frequency sub-band coefficient matrix H 'kWavelet inverse transformation is done with other frequency domain sub-bands in original picture block, obtains insertion water
Image block A ' after printkj；
(9) by each piece of A 'kjReplace its original piecemeal in R, G, B component, obtain R, G containing watermark, after B component synthesis contain
Watermarking images OW。
3. a kind of be embedded in based on the Watermarking for Color Images of compressed sensing and wavelet transformation according to claim 1 is extracted
Method, which is characterized in that the step of digital watermarking extracting method is as follows:
(1) by aqueous printed color picture OWIt carries out R, G, B to decompose, size homalographic when respectively to R, G, B component by watermark insertion
Non-overlap piecemeal chooses piecemeal according to key key2 and constitutes insertion arrayK=(R, G, B)；
(2) by arrayIn every piece of progress wavelet decomposition, extract high frequency coefficient matrix after, sparse basis Ψ is obtained by key3kAnd sight
Survey matrix Φk, every piece of measured value is obtained after compressed sensing processing:
(3) the measured value Y ' of watermark blockWi:
αiValue can be obtained by key4；
(4) compressed sensing reconstruct is carried out to all watermark block measured values of extraction, synthesizes watermarking images W ' after obtaining watermarking images block.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910608251.0A CN110428355A (en) | 2019-07-08 | 2019-07-08 | A kind of Watermarking for Color Images method based on compressed sensing and NSCT |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910608251.0A CN110428355A (en) | 2019-07-08 | 2019-07-08 | A kind of Watermarking for Color Images method based on compressed sensing and NSCT |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110428355A true CN110428355A (en) | 2019-11-08 |
Family
ID=68410324
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910608251.0A Pending CN110428355A (en) | 2019-07-08 | 2019-07-08 | A kind of Watermarking for Color Images method based on compressed sensing and NSCT |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110428355A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111340675A (en) * | 2020-02-12 | 2020-06-26 | 中南林业科技大学 | Colored pattern watermark embedding and extracting method based on sparse representation |
| CN111681153A (en) * | 2020-06-08 | 2020-09-18 | 重庆邮电大学 | CS-SVD robust digital watermarking algorithm based on non-subsampled contourlet transform |
-
2019
- 2019-07-08 CN CN201910608251.0A patent/CN110428355A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111340675A (en) * | 2020-02-12 | 2020-06-26 | 中南林业科技大学 | Colored pattern watermark embedding and extracting method based on sparse representation |
| CN111340675B (en) * | 2020-02-12 | 2023-04-25 | 中南林业科技大学 | Sparse representation-based color pattern watermark embedding and extracting method |
| CN111681153A (en) * | 2020-06-08 | 2020-09-18 | 重庆邮电大学 | CS-SVD robust digital watermarking algorithm based on non-subsampled contourlet transform |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106023056B (en) | Zero watermarking insertion, extracting method and the device compressed based on DWT and principal component analysis | |
| CN102750660B (en) | A kind of method and apparatus embedding and extract digital watermarking | |
| Chang et al. | Features classification forest: a novel development that is adaptable to robust blind watermarking techniques | |
| Divecha et al. | Implementation and performance analysis of DCT-DWT-SVD based watermarking algorithms for color images | |
| CN104053074B (en) | Video watermarking method based on depth image and Otsu segmentation | |
| CN108564519A (en) | A kind of Watermarking for Color Images method based on different transform domain | |
| CN101246588B (en) | Self-adapting watermarking algorithm of colorful image hypercomplex number spacing | |
| Jain et al. | A reliable SVD based watermarking schem | |
| CN105512999B (en) | A kind of color image holographic watermark method of double transformation | |
| CN113989096B (en) | Robust image watermarking method and system based on deep learning and attention network | |
| Jie | A novel block-DCT and PCA based image perceptual hashing algorithm | |
| CN105787863A (en) | Linear canonical wavelet transform (LCWT)-based digital watermarking method | |
| Ramakrishnan et al. | Svd based robust digital watermarking for still images using wavelet transform | |
| CN106204410B (en) | A kind of novel digital watermark method decomposed based on matrix Schur | |
| CN104835106B (en) | A kind of full frequency-domain subband data waterprint embedded method based on wavelet decomposition | |
| CN110189243A (en) | A kind of color image robust watermarking method based on tensor singular value decomposition | |
| Garg et al. | Optimized color image watermarking through watermark strength optimization using particle swarm optimization technique | |
| CN110428355A (en) | A kind of Watermarking for Color Images method based on compressed sensing and NSCT | |
| Kekre et al. | Robust Watermaking Technique Using Hybrid Wavelet Transform Generated From Kekre Transforms and Discrete Cosine Transform | |
| Yuan et al. | A multiscale fragile watermark based on the Gaussian mixture model in the wavelet domain | |
| Bansal et al. | Performance evaluation of steganography tools using SVM and NPR tool | |
| Hu et al. | A blind watermarking algorithm for color image based on wavelet transform and Fourier transform | |
| Asatryan et al. | Combined spatial and frequency domain watermarking | |
| CN109410115A (en) | The insertion of adaptive capacity image blind watermarking and extracting method based on SIFT feature | |
| Silja et al. | A watermarking algorithm based on contourlet transform and nonnegative matrix factorization |
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 |