CN107507122A - Stereo-picture Zero watermarking method based on NSCT and SIFT - Google Patents
Stereo-picture Zero watermarking method based on NSCT and SIFT Download PDFInfo
- Publication number
- CN107507122A CN107507122A CN201710722220.9A CN201710722220A CN107507122A CN 107507122 A CN107507122 A CN 107507122A CN 201710722220 A CN201710722220 A CN 201710722220A CN 107507122 A CN107507122 A CN 107507122A
- Authority
- CN
- China
- Prior art keywords
- mrow
- image
- msub
- mtd
- msup
- 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
- 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
- G06T1/0064—Geometric transfor invariant watermarking, e.g. affine transform invariant
-
- 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
- G06T1/0085—Time domain based watermarking, e.g. watermarks spread over several images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2201/00—General purpose image data processing
- G06T2201/005—Image watermarking
- G06T2201/0051—Embedding of the watermark in the spatial domain
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
Abstract
Stereo-picture Zero watermarking method category image encryption and technical field of image processing based on NSCT and SIFT; the low frequency subband image of the left and right visual point image NSCT transform domains of present invention extraction stereo-picture; carrier using the anaglyph between them as extraction watermark; singular value decomposition is carried out to carrier image; formulate the rule of extraction zero watermarking; certification zero watermarking image is obtained after the two-value zero watermarking image extracted and copyright marking watermarking images are carried out into XOR; after registration center preserves certification zero watermarking image, stereo-picture has just been in the protection of digital watermark.The present invention can resist the conventional attacks such as various noises, filtering, compression well; it has been improved particularly resisting the ability of geometric attack; zero watermarking insertion does not change original image; it can guarantee that the quality of image; and original image is not needed during watermark extracting; provided convenience for practical application, can effectively protect the copyright of stereo-picture.
Description
Technical field
The invention belongs to image encryption and technical field of image processing, and in particular to a kind of zero watermarking side of stereo-picture
Method.
Background technology
With the continuous development of multimedia technology, traditional single view plane picture can not meet the needs of people
, for the scene information that one-view image provides than relatively limited, visual experience is also not strong enough, and can provide the vertical of depth information
Body image can get up the information integration on different directions and different angle, give sensation on the spot in person, enhance pair
The degree of awareness of object.Stereo-picture technology is applied at many aspects, for example, 3D films, teleconference, medical treatment,
Military affairs etc..Stereo-picture technology has good development prospect and application.
But due to the fast development of internet, image is easy to be transmitted, and steals and distorts, thing followed stereogram
The safety problem of picture is urgently to be resolved hurrily.Digital watermarking is a kind of effective copyright protection technology developed in recent years, turns into letter
Cease the hot technology of security fields research.Digital watermarking utilizes the intrinsic characteristic of image-carrier, and watermark information is embedded into carrier
In, judge the ownership of the authenticity of picture material and copyright by detecting the integrity degree of watermark information in receiving terminal.Watermark skill
Art is disguised preferably, the use of image will not be impacted, be a kind of means for being effectively protected image copyright.
At present, the research for two-dimensional image water mark method is very deep, technology also relative maturity, is broadly divided into
Spatial domain method and frequency domain method.Spatial domain method is simple and easy, but the robustness of watermark is poor, it is difficult to resisting various attacks;And frequency
Domain method realizes the insertion of watermark generally by the coefficient of modification frequency-domain transform, and the change of frequency coefficient can be distributed to each
The pixel in spatial domain, the robustness of watermark are also just improved.But the left and right that stereo-picture is arrived by two-way camera acquisition
Visual point image is respectively fed in the right and left eyes of people, is formed binocular parallax, real scene is fused into human brain, compared to plan
As, it is necessary to consider the contact between more many factors, such as left and right visual point image, it is impossible to simply by plane picture watermark
Method is applied on stereo-picture.Also, traditional spatial domain and transform domain method can all cause to the picture quality of stereo-picture
Influence, it is possible to destroy stereoscopic visual effect.
Research on stereo-picture water mark method is fewer, and existing method can not take into account resistance routine completely and attack
Hit and geometric attack, and ensure that picture quality is unaffected.
The content of the invention
It is an object of the invention to provide one kind, based on NSCT (non-down sampling contourlet transform) and SIFT, (scale feature is not
Become conversion) stereo image parallax Zero watermarking method, strong robustness of the present invention, original image quality is not influenceed, practicality
By force, universality is good.
A kind of stereo-picture Zero watermarking method based on NSCT and SIFT of the present invention, including watermark telescopiny and watermark
Extraction process;
1. watermark telescopiny comprises the following steps:
The database of 1.1 selection stereo-pictures, chooses the stereo pairs for emulation, by the right viewpoint figure of stereo-picture
As being denoted as R, left view dot image is denoted as L, and two images size is all M × N, and two-value copyright marking watermarking images are denoted as WL, size
For m × n;
1.2 by left and right visual point image gray processing, carries out SIFT feature extraction to two images respectively:
1) when Gaussian difference scale space extracts SIFT feature, image I (x, y) metric space L (x, y, δ) is under
Row formula (1) and formula (2) obtain:
L (x, y, δ)=I (x, y) * G (x, y, δ) (1)
Wherein:G (x, y, δ) is gaussian kernel function, and (x, y) is space pixel coordinate, and δ is yardstick coordinate, thus obtains height
This metric space:
2) after obtaining metric space, 8 points of each point with adjacent same yardstick, and 18 points of neighbouring yardstick
It is compared, finds out extreme point;All extreme points are screened using Hessian matrix, Hessian matrix and stability formula
As shown in formula (4) and formula (5):
Wherein:D is differential operator;Tr (H) is matrix H diagonal entry sum;Det (H) is determinant of a matrix;E is
The ratio of characteristic point eigenvalue of maximum and minimal eigenvalue;If the extreme point found out meets formula (4) and formula (5), so that it may
To be designated as stable characteristic point;If the collection of stable characteristic point is combined into:
F={ fi|fi=((xi,yi),si,θi),i∈(0,p)} (6)
Wherein:(xi, yi) it is characterized location of pixels a little;siFor the characteristic dimension of image;θiIt is characterized principal direction a little;p
It is characterized number a little;
3) 128 dimensional feature point descriptors are generated;The gradient magnitude of 16 × 16 matrix dot near characteristic point is calculated first
With gradient angle, the feature point description symbol of one 4 × 4 is then formed, totally 16, each includes 8 histogram directions, totally 4 × 4
× 8=128 vector, you can represent the descriptor of characteristic point;
1.3 couples of left view dot image L and right visual point image R carry out non-down sampling contourlet NSCT decomposition respectively, obtain low frequency
Sub-band images SLAnd SR, and the high-frequency sub-band images on each yardstick;
1.4 take out low frequency sub-band SLAnd SR, using the method for Stereo image matching, ask between two width low frequency subband images
Anaglyph D, using the disparity map in horizontal direction, and matching way from left to right, the size of anaglyph with it is original
Stereo-picture is identical, is also M × N;
1.5 are divided into anaglyph nonoverlapping image subblock C that size is (M/m) × (N/n)i, of image subblock
Number is m × n, and serial number, 1≤i≤m × n are demarcated according to order from left to right from top to bottom to image subblock with i;
1.6 pairs of each image subblocks carry out singular value decomposition SVD, as shown in formula (7):
Wherein:U and V is unitary matrice;S is diagonal matrix, i.e. singular value matrix, and the element on diagonal is singular value, S=
diag{λ1, λ2, λ3,……λm};Take out each first eigenvalue λ of image subblock singular value1i, that is, in matrix it is maximum
Characteristic value, seek all λ1iAverage value λa, 1≤i≤m × n, such as formula (8):
1.7 construct zero watermarking according to the singular value of each image subblock and the relation of their average value, formulate zero water
The create-rule of print:If the singular value of i-th of image subblock is more than average value, w is setiFor 1;If i-th of image subblock
Singular value be less than average value, then set wiFor 0, such as formula (9);All image subblocks are traveled through in order, all w that will be obtainedi
A binary sequence is accumulated, binary sequence is deformed into the two-value zero watermarking image W that size is m × n;
1.8 by copyright marking watermarking images WLXOR is carried out with two-value zero watermarking image W, obtains authenticating water-mark image W*,
Authenticating water-mark image is sent in the watermark information storehouse of registration center and preserved, make stereo-picture be in digital watermark protection it
In;
Preserved the SIFT feature template of the generating mode of zero watermarking, create-rule and original image as key;
2 watermark extraction process comprise the following steps:
The left and right visual point image of 2.1 pairs of stereo-pictures to be detected is denoted as L' and R' respectively, and size is M × N, with preservation
Feature templates carry out SIFT feature matching first, judge whether by geometric attack, if there is geometric attack, carry out image calibration
Just, otherwise, directly carry out in next step;
2.2 will carry out NSCT decomposition respectively after left view dot image L' and right visual point image R' gray processings, obtain low frequency sub-band
With high-frequency sub-band, respective low frequency subband image S' is taken outLAnd S'R, size is M × N;
2.3 methods that two width low frequency subband images are utilized to Stereo matching, obtain the anaglyph D' in horizontal direction,
It is from left to right with mode;
2.4 resolve into anaglyph nonoverlapping sub-block C' that size is (M/m) × (N/n)i, by image subblock according to
From left to right order from top to bottom is arranged, with the order of i representative image sub-blocks, 1≤i≤m × n;
2.5 carry out singular value decomposition to each image subblock in order:
First singular value λ ' is extracted in each singular value matrixi1, ask for their average value:
The relation of 2.6 traversals all image subblock, the successively singular value of more each image subblock and their singular values,
Such as formula (13), if i-th of singular value is bigger than average value, return value 1, otherwise, return value 0, by all return values in order
It is stored in the two values matrix that size is m × n, the matrix is exactly the two-value zero watermarking image W' extracted;
2.7 take out the authenticating water-mark image W* preserved from registration center, with the two-value zero watermarking figure extracted in 1.2.6 steps
As W' progress XORs, copyright mark watermark will image W' is obtainedL:
Differentiate the attaching problem of copyright with the content of copyright marking watermarking images.
Compared with prior art, it is an advantage of the invention that:
1) construct parallax zero watermarking using the robustness of the singular value of stereo-picture NSCT transform domain low frequency sub-bands, when by
To remaining able to extract watermark among image exactly during various attacks, compared to other method, the Shandong of watermark is improved
Rod, while the redundancy of NSCT conversion also improves the capacity of watermark.
2) watermark is constructed by the use of anaglyph as carrier, can guarantee that contacting between watermark and left and right visual point image.
3) solves the contradiction between water mark method robustness and imperceptibility, zero watermarking telescopiny will not be to original
Image causes to change, and can guarantee that the picture quality of original image.
4) ability of watermark resistance geometric attack can be improved, geometric attack can destroy synchronous between watermark information and image
Property, using SIFT feature Point matching watermark will be extracted again after image rectification under attack.
5) method proposed by the invention is blind watermark method, does not need original image when extracting watermark, is practical application
Provide convenience.
Brief description of the drawings
Fig. 1 is the flow chart of the stereo-picture zero watermarking based on NSCT and SIFT
Fig. 2 is stereo-picture Dolls, and (a) is left view dot image, and (b) is right visual point image
Fig. 3 is copyright marking watermarking images
Fig. 4 is stereo-picture Teddy, and (a) is left view dot image, and (b) is right visual point image
Fig. 5 be stereo-picture by the watermark extracted after asymmetrical attack, (a)~(j) be followed successively by by 10 kinds attack
Copyright marking watermarking images are extracted after type
Fig. 6 is the watermark extracted after stereo-picture is symmetrically attacked, and (a)~(j) is followed successively by by 10 kinds of attack classes
The copyright marking watermarking images extracted after type
Embodiment
The embodiment of the present invention is described below in conjunction with the accompanying drawings, so as to those skilled in the art preferably
Understand the present invention.
As shown in figure 1, the watermark telescopiny of the present invention, comprises the following steps:
1. selecting the database of stereo-picture, Fig. 2 is left view dot image and the right side for choosing the stereo pairs for emulating
Visual point image, size are 512 × 512;Fig. 3 is two-value copyright marking watermarking images, is denoted as WL, size is 64 × 64.
1.2. by left and right visual point image gray processing, SIFT feature extraction is carried out to two images respectively:
It is at Gaussian difference scale space (Difference ofGaussian, DOG) when 1.2.1 extracting SIFT feature
Carry out, image I (x, y) metric space L (x, y, δ) can be obtained by formula (15) and formula (16):
L (x, y, δ)=I (x, y) * G (x, y, δ) (15)
Wherein, G (x, y, δ) is gaussian kernel function;(x, y) is space pixel coordinate;δ is yardstick coordinate, it can thus be concluded that arriving
Gaussian scale-space:
1.2.2 after obtaining metric space, 8 points of each point with adjacent same yardstick, and the 18 of neighbouring yardstick
Individual point is compared, and finds out extreme point;All extreme points are screened using Hessian matrix, Hessian matrix and stable calculation
Shown in formula such as formula (18) and formula (19):
Wherein:Wherein, D is differential operator;Tr (H) is matrix H diagonal entry sum;Det (H) is the ranks of matrix
Formula;E is characterized the ratio of an eigenvalue of maximum and minimal eigenvalue;If the extreme point found out meets formula (18) and public affairs above
Formula (19), it is possible to be designated as stable characteristic point;If the collection of stable characteristic point is combined into
F={ fi|fi=((xi,yi),si,θi),i∈(0,p)} (20)
Wherein:(xi,yi) it is characterized location of pixels a little;siFor the characteristic dimension of image;θiIt is characterized principal direction a little, p
It is characterized number a little.
1.2.3 128 dimensional feature point descriptors are generated;Calculate characteristic point near 16 × 16 matrix dot gradient magnitude and
Gradient angle, then form the feature point description symbol of one 4 × 4, totally 16, each include 8 histogram directions, i.e., totally 4 × 4
× 8=128 vector, you can represent the descriptor of characteristic point.
1.2.4 the position of left view dot image and the respective characteristic point of right visual point image, yardstick, description are saved as into template
TLAnd TR。
1.3. non-down sampling contourlet NSCT decomposition is carried out respectively to left view dot image L and right visual point image R, obtains low frequency
Sub-band images SLAnd SR, and the high-frequency sub-band images on each yardstick.
1.4. low frequency sub-band S is taken outLAnd SR, using the matching process of stereo-picture, ask between two width low frequency subband images
Anaglyph D, here we using disparity map in the horizontal direction, matching way is from left to right anaglyph
Size it is identical with original three-dimensional image, be also 512 × 512.
1.5. anaglyph is divided into nonoverlapping image subblock C that size is 8 × 8i, the number of image subblock for 64 ×
64, with i to image subblock according to sequential calibration order sequence number from left to right from top to bottom, 1≤i≤64 × 64.
1.6. singular value decomposition (SVD) is carried out to each image subblock:
Wherein, U and V is unitary matrice;S is diagonal matrix, i.e. singular value matrix, and the element on diagonal is singular value, S=
diag{λ1, λ2, λ3,……λm, there are some researches show for the singular value of image when being interfered and attacking, robustness is very strong;Take
Go out each first eigenvalue λ of image subblock singular value1i, that is, characteristic value maximum in matrix, seek all λ1iBe averaged
Value λa, 1≤i≤m × n, as shown in formula (22):
1.7. zero watermarking is constructed according to the singular value of each image subblock and the relation of their average value, formulates zero water
The create-rule of print:If the singular value of i-th of image subblock is more than average value, w is setiFor 1;If i-th of image subblock
Singular value be less than average value, then set wiFor 0, such as formula (23);All image subblocks are traveled through in order, it is all by what is obtained
wiA binary sequence is accumulated, binary sequence is deformed into the two-value zero watermarking image W that size is 64 × 64.
1.8. by copyright marking watermarking images WLXOR is carried out with two-value zero watermarking image W, authenticating water-mark W* is obtained, will recognize
Card watermark, which is sent in the watermark information storehouse of registration center, to be preserved, and stereo-picture is among the protection of digital watermark.
Preserved the SIFT feature template of the generating mode of zero watermarking, create-rule and original image as key.
As shown in Fig. 1 right parts, watermark extraction process of the invention, comprise the following steps:
2.1 are attacked stereo-picture with different methods, to the left and right visual point image of stereo pairs to be detected
Be denoted as L' and R' respectively, size is 512 × 512, and SIFT feature matching is carried out first with the feature templates of preservation, judge whether by
To geometric attack, if there is geometric attack, image rectification is carried out, otherwise, is directly carried out in next step.
2.2 will carry out NSCT decomposition respectively after left view dot image L' and right visual point image R' gray processings, obtain low frequency sub-band
With high-frequency sub-band, respective low frequency sub-band S' is taken outLAnd S'R。
2.3 methods that low frequency subband image is utilized to Stereo matching, obtain the anaglyph D' in horizontal direction, match party
Formula is from left to right.
2.4 resolve into anaglyph nonoverlapping sub-block C' that size is 8 × 8i, by image subblock according to from left to right
Order from top to bottom is arranged, with i come the order of representative image sub-block, 1≤i≤64 × 64.
2.5 carry out singular value decomposition to each image subblock in order:
First singular value λ ' is extracted in each singular value matrixi1, ask for their average value:
The relation of 2.6 traversals all image subblock, the successively singular value of more each image subblock and their singular values,
If formula (27) is if i-th of singular value is bigger than average value, return value 1, otherwise, return value 0.All return values are deposited in order
Store up in the two values matrix that size is 64 × 64, the matrix is exactly the two-value zero watermarking image W' extracted.
2.7 take out the authenticating water-mark image W* preserved from registration center, with the two-value zero watermarking image extracted in 2.6 steps
W' carries out XOR, obtains copyright marking watermarking images W'L:
As shown in Figure 5,6, the content of copyright marking watermarking images can prove copyright ownership.
In order to weigh the validity of water mark method, Y-PSNR (PSNR) is used for weighing the quality of image, normalizing
Change the water that coefficient correlation (NC) is used for comparing the watermark extracted among original image with extracting among image pair under attack
Similarity degree between print.PSNR is defined as follows:
Wherein:I (x, y) represents pixel value of the original image in (x, y) point;Iw(x, y) represent it is under attack after image
In the pixel value of (x, y) point, the size of image is all M × N;The peak value of D representative image signals, is traditionally arranged to be 255.PSNR's
Numerical value is lower, and the quality of representative image is poorer.
NC is defined as:
Wherein:W (i, j) is the pixel value of the watermark that extracts among original image in (i, j) point;W'(i, j) be by
Pixel value of the watermark extracted among the image of attack in (i, j) point;The size of watermark is m × n;NC values are bigger, it is meant that two
Width watermarking images are more close, that is, water mark method attack tolerant can be better.
The construction process of zero watermarking is relevant with the inwardness of stereo-picture, so the water that different stereo-pictures are extracted
Print should be different.The similarity degree between the different zero watermarkings of extraction is weighed with the size of NC values.Fig. 2 extract zero
NC values between the zero watermarking that watermark and Fig. 4 are extracted are 0.5882, and much smaller than 1, correlation is very low, so the present invention can protect
The uniqueness of watermark between card different images.
Various types of attacks are carried out to the image of embedded watermark below, verify effectiveness of the invention.Attack type point
For asymmetrical attack and symmetrical attack.Asymmetrical attack is only to attack right visual point image, and left view dot image keeps constant, symmetrically
Attack is to attack left and right visual point image simultaneously.Table 1 be by after various attacks, the left and right visual point image of Fig. 2 neutral body images
PSNR values, and the zero watermarking that is extracted from original image and the zero watermarking that watermark is extracted from image under attack
Between NC values.
Table 1
Can be seen that the present invention from the data in table 1 can be effective against various conventional attacks, and it is several especially to lift resistance
The ability of what attack, and the insertion of watermark does not interfere with the quality of image, and the copyright marking watermarking images extracted are effective
Demonstrate copyright ownership.As shown in Figure 5 and Figure 6, from visual results as can be seen that even if image has been seriously damaged, still
Clearly copyright marking watermarking images can successfully be extracted and confirm image ownership.
The present invention consider while water mark method robustness strengthen, do not influence the quality of carrier image, using NSCT with
SVD robustness constructs parallax zero watermarking, and zero watermarking will not be modified in telescopiny to carrier image, introduces SIFT
The ability of feature point calibration enhancing watermark resistance geometric attack, NSCT redundancy also enhance the ability of resistance attacked by noise,
And watermark capacity is improved, original image is not needed in watermark extraction process, practical application facility is enhanced, realizes pair
The copyright protection of stereo-picture and certification.
Claims (1)
1. a kind of stereo-picture Zero watermarking method based on NSCT and SIFT, it is characterised in that including watermark telescopiny and watermark
Extraction process;
1.1 watermark telescopinies comprise the following steps:
1.1.1 the database of stereo-picture is selected, the stereo pairs for emulation are chosen, by the right visual point image of stereo-picture
R is denoted as, left view dot image is denoted as L, and two images size is all M × N, and two-value copyright marking watermarking images are denoted as WL, size m
×n;
1.1.2 by left and right visual point image gray processing, SIFT feature extraction is carried out to two images respectively:
1) when Gaussian difference scale space extracts SIFT feature, image I (x, y) metric space L (x, y, δ) is by following public affairs
Formula (1) and formula (2) obtain:
L (x, y, δ)=I (x, y) * G (x, y, δ) (1)
<mrow>
<mi>G</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>,</mo>
<mi>&delta;</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>2</mn>
<msup>
<mi>&pi;&delta;</mi>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mi>exp</mi>
<mrow>
<mo>(</mo>
<mo>-</mo>
<mfrac>
<mrow>
<msup>
<mi>x</mi>
<mn>2</mn>
</msup>
<mo>+</mo>
<msup>
<mi>y</mi>
<mn>2</mn>
</msup>
</mrow>
<mrow>
<mn>2</mn>
<msup>
<mi>&delta;</mi>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein:G (x, y, δ) is gaussian kernel function, and (x, y) is space pixel coordinate, and δ is yardstick coordinate, thus obtains Gauss chi
Spend space:
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<mi>D</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>,</mo>
<mi>&delta;</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mrow>
<mo>(</mo>
<mi>G</mi>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>,</mo>
<mi>k</mi>
<mi>&delta;</mi>
</mrow>
<mo>)</mo>
<mo>-</mo>
<mi>G</mi>
<mo>(</mo>
<mrow>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>,</mo>
<mi>&delta;</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>*</mo>
<mi>I</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>=</mo>
<mi>L</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>,</mo>
<mi>k</mi>
<mi>&delta;</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mi>L</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>,</mo>
<mi>&delta;</mi>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
</mrow>
2) after obtaining metric space, 8 points of each point with adjacent same yardstick, and 18 points of neighbouring yardstick are carried out
Compare, find out extreme point;All extreme points are screened using Hessian matrix, Hessian matrix and stability formula such as formula
(4) and shown in formula (5):
<mrow>
<mi>H</mi>
<mo>=</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<msub>
<mi>D</mi>
<mrow>
<mi>x</mi>
<mi>x</mi>
</mrow>
</msub>
</mtd>
<mtd>
<msub>
<mi>D</mi>
<mrow>
<mi>x</mi>
<mi>y</mi>
</mrow>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>D</mi>
<mrow>
<mi>x</mi>
<mi>y</mi>
</mrow>
</msub>
</mtd>
<mtd>
<msub>
<mi>D</mi>
<mrow>
<mi>y</mi>
<mi>y</mi>
</mrow>
</msub>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>s</mi>
<mo>=</mo>
<mfrac>
<mrow>
<mi>T</mi>
<mi>r</mi>
<msup>
<mrow>
<mo>(</mo>
<mi>H</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
<mrow>
<mi>D</mi>
<mi>e</mi>
<mi>t</mi>
<mrow>
<mo>(</mo>
<mi>H</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>=</mo>
<mfrac>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>D</mi>
<mrow>
<mi>x</mi>
<mi>x</mi>
</mrow>
</msub>
<mo>+</mo>
<msub>
<mi>D</mi>
<mrow>
<mi>y</mi>
<mi>y</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mrow>
<mo>(</mo>
<msub>
<mi>D</mi>
<mrow>
<mi>x</mi>
<mi>x</mi>
</mrow>
</msub>
<msub>
<mi>D</mi>
<mrow>
<mi>y</mi>
<mi>y</mi>
</mrow>
</msub>
<mo>-</mo>
<msup>
<msub>
<mi>D</mi>
<mrow>
<mi>x</mi>
<mi>y</mi>
</mrow>
</msub>
<mn>2</mn>
</msup>
<mo>)</mo>
</mrow>
</mfrac>
<mo><</mo>
<mfrac>
<msup>
<mrow>
<mo>(</mo>
<mi>e</mi>
<mo>+</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mi>e</mi>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>5</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein:D is differential operator;Tr (H) is matrix H diagonal entry sum;Det (H) is determinant of a matrix;E is characterized
The ratio of point eigenvalue of maximum and minimal eigenvalue;If the extreme point found out meets formula (4) and formula (5), it is possible to remembers
For stable characteristic point;If the collection of stable characteristic point is combined into:
F={ fi|fi=((xi,yi),si,θi),i∈(0,p)} (6)
Wherein:(xi, yi) it is characterized location of pixels a little;siFor the characteristic dimension of image;θiIt is characterized principal direction a little;P is spy
Levy the number of point;
3) 128 dimensional feature point descriptors are generated;The gradient magnitude and ladder of 16 × 16 matrix dot near characteristic point are calculated first
Angle is spent, the feature point description symbol of one 4 × 4 is then formed, totally 16, each includes 8 histogram directions, totally 4 × 4 × 8
=128 vectors, you can represent the descriptor of characteristic point;
1.1.3. non-down sampling contourlet NSCT decomposition is carried out respectively to left view dot image L and right visual point image R, obtains low frequency
Band image SLAnd SR, and the high-frequency sub-band images on each yardstick;
1.1.4. take out low frequency sub-band SLAnd SR, using the method for Stereo image matching, ask between two width low frequency subband images
Anaglyph D, using the disparity map in horizontal direction, and matching way from left to right, the size of anaglyph are stood with original
Body image is identical, is also M × N;
1.1.5. anaglyph is divided into nonoverlapping image subblock C that size is (M/m) × (N/n)i, the number of image subblock
For m × n, serial number, 1≤i≤m × n are demarcated according to order from left to right from top to bottom to image subblock with i;
1.1.6. singular value decomposition SVD is carried out to each image subblock, as shown in formula (7):
<mrow>
<msub>
<mi>C</mi>
<mi>i</mi>
</msub>
<mo>=</mo>
<msub>
<mi>U</mi>
<mrow>
<mi>M</mi>
<mo>&times;</mo>
<mi>M</mi>
</mrow>
</msub>
<msub>
<mi>S</mi>
<mrow>
<mi>M</mi>
<mo>&times;</mo>
<mi>N</mi>
</mrow>
</msub>
<msubsup>
<mi>V</mi>
<mrow>
<mi>N</mi>
<mo>&times;</mo>
<mi>N</mi>
</mrow>
<mi>T</mi>
</msubsup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>7</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein:U and V is unitary matrice;S is diagonal matrix, i.e. singular value matrix, and the element on diagonal is singular value, S=diag
{λ1, λ2, λ3,……λm};Take out each first eigenvalue λ of image subblock singular value1i, that is, feature maximum in matrix
Value, seeks all λ1iAverage value λa, 1≤i≤m × n, such as formula (8):
<mrow>
<msub>
<mi>&lambda;</mi>
<mi>a</mi>
</msub>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<mi>m</mi>
<mo>&times;</mo>
<mi>n</mi>
</mrow>
</munderover>
<msub>
<mi>&lambda;</mi>
<mrow>
<mn>1</mn>
<mi>i</mi>
</mrow>
</msub>
<mo>/</mo>
<mrow>
<mo>(</mo>
<mi>m</mi>
<mo>&times;</mo>
<mi>n</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>8</mn>
<mo>)</mo>
</mrow>
</mrow>
1.1.7. zero watermarking is constructed according to the singular value of each image subblock and the relation of their average value, formulates zero watermarking
Create-rule:If the singular value of i-th of image subblock is more than average value, w is setiFor 1;If i-th image subblock
Singular value is less than average value, then sets wiFor 0, such as formula (9);All image subblocks are traveled through in order, all w that will be obtainediConverge
A binary sequence is integrated, binary sequence is deformed into the two-value zero watermarking image W that size is m × n;
<mrow>
<msub>
<mi>w</mi>
<mi>i</mi>
</msub>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>&lambda;</mi>
<mi>a</mi>
</msub>
<mo>></mo>
<msub>
<mi>&lambda;</mi>
<mi>i</mi>
</msub>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>&lambda;</mi>
<mi>a</mi>
</msub>
<mo><</mo>
<msub>
<mi>&lambda;</mi>
<mi>i</mi>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>9</mn>
<mo>)</mo>
</mrow>
</mrow>
1.1.8. by copyright marking watermarking images WLXOR is carried out with two-value zero watermarking image W, obtains authenticating water-mark image W*, will
Authenticating water-mark image is sent in the watermark information storehouse of registration center and preserved, and stereo-picture is among the protection of digital watermark;
<mrow>
<msup>
<mi>W</mi>
<mo>*</mo>
</msup>
<mo>=</mo>
<mi>W</mi>
<mo>&CirclePlus;</mo>
<msub>
<mi>W</mi>
<mi>L</mi>
</msub>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>10</mn>
<mo>)</mo>
</mrow>
</mrow>
Preserved the SIFT feature template of the generating mode of zero watermarking, create-rule and original image as key;
1.2 watermark extraction process comprise the following steps:
1.2.1 L' and R' are denoted as respectively to the left and right visual point image of stereo-picture to be detected, size is M × N, with the spy of preservation
Sign template carries out SIFT feature matching first, judges whether by geometric attack, if there is geometric attack, carries out image calibration
Just, otherwise, directly carry out in next step;
1.2.2 NSCT decomposition will be carried out respectively after left view dot image L' and right visual point image R' gray processings, obtain low frequency sub-band with
High-frequency sub-band, take out respective low frequency subband image S'LAnd S'R, size is M × N;
1.2.3, the method that two width low frequency subband images are utilized to Stereo matching, obtains the anaglyph D' in horizontal direction, matches
Mode is from left to right;
1.2.4 anaglyph is resolved into nonoverlapping sub-block C' that size is (M/m) × (N/n)i, by image subblock according to from
Left-to-right order from top to bottom is arranged, with the order of i representative image sub-blocks, 1≤i≤m × n;
1.2.5 singular value decomposition is carried out to each image subblock in order:
<mrow>
<msup>
<msub>
<mi>C</mi>
<mi>i</mi>
</msub>
<mo>&prime;</mo>
</msup>
<mo>=</mo>
<msub>
<msup>
<mi>U</mi>
<mo>&prime;</mo>
</msup>
<mrow>
<mi>M</mi>
<mo>&times;</mo>
<mi>M</mi>
</mrow>
</msub>
<msub>
<msup>
<mi>S</mi>
<mo>&prime;</mo>
</msup>
<mrow>
<mi>M</mi>
<mo>&times;</mo>
<mi>N</mi>
</mrow>
</msub>
<msubsup>
<mi>V</mi>
<mrow>
<mi>N</mi>
<mo>&times;</mo>
<mi>N</mi>
</mrow>
<mrow>
<mo>&prime;</mo>
<mi>T</mi>
</mrow>
</msubsup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>11</mn>
<mo>)</mo>
</mrow>
</mrow>
First singular value λ ' is extracted in each singular value matrixi1, ask for their average value:
<mrow>
<msubsup>
<mi>&lambda;</mi>
<mi>a</mi>
<mo>&prime;</mo>
</msubsup>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<mi>m</mi>
<mo>&times;</mo>
<mi>n</mi>
</mrow>
</munderover>
<msubsup>
<mi>&lambda;</mi>
<mrow>
<mn>1</mn>
<mi>i</mi>
</mrow>
<mo>&prime;</mo>
</msubsup>
<mo>/</mo>
<mrow>
<mo>(</mo>
<mi>m</mi>
<mo>&times;</mo>
<mi>n</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>12</mn>
<mo>)</mo>
</mrow>
</mrow>
1.2.6 all image subblocks are traveled through, successively the relation of the singular value of more each image subblock and their singular values, such as
Formula (13), if i-th of singular value is bigger than average value, return value 1, otherwise, return value 0, all return values are deposited in order
Store up in the two values matrix that size is m × n, the matrix is exactly the two-value zero watermarking image W' extracted;
<mrow>
<msubsup>
<mi>w</mi>
<mi>i</mi>
<mo>&prime;</mo>
</msubsup>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msubsup>
<mi>&lambda;</mi>
<mi>a</mi>
<mo>&prime;</mo>
</msubsup>
<mo>></mo>
<msubsup>
<mi>&lambda;</mi>
<mrow>
<mn>1</mn>
<mi>i</mi>
</mrow>
<mo>&prime;</mo>
</msubsup>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msubsup>
<mi>&lambda;</mi>
<mi>a</mi>
<mo>&prime;</mo>
</msubsup>
<mo><</mo>
<msubsup>
<mi>&lambda;</mi>
<mrow>
<mn>1</mn>
<mi>i</mi>
</mrow>
<mo>&prime;</mo>
</msubsup>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>13</mn>
<mo>)</mo>
</mrow>
</mrow>
1.2.7 the authenticating water-mark image W* preserved is taken out from registration center, with the two-value zero watermarking image extracted in 1.2.6 steps
W' carries out XOR, obtains copyright mark watermark will image W'L:
<mrow>
<msubsup>
<mi>W</mi>
<mi>L</mi>
<mo>&prime;</mo>
</msubsup>
<mo>=</mo>
<msup>
<mi>W</mi>
<mo>&prime;</mo>
</msup>
<mo>&CirclePlus;</mo>
<msup>
<mi>W</mi>
<mo>*</mo>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>14</mn>
<mo>)</mo>
</mrow>
</mrow>
Differentiate the attaching problem of copyright with the content of copyright marking watermarking images.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710722220.9A CN107507122B (en) | 2017-08-22 | 2017-08-22 | Stereo image zero-watermarking method based on NSCT and SIFT |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710722220.9A CN107507122B (en) | 2017-08-22 | 2017-08-22 | Stereo image zero-watermarking method based on NSCT and SIFT |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107507122A true CN107507122A (en) | 2017-12-22 |
CN107507122B CN107507122B (en) | 2020-10-16 |
Family
ID=60692451
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710722220.9A Active CN107507122B (en) | 2017-08-22 | 2017-08-22 | Stereo image zero-watermarking method based on NSCT and SIFT |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107507122B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109410115A (en) * | 2018-10-31 | 2019-03-01 | 山东省计算中心(国家超级计算济南中心) | The insertion of adaptive capacity image blind watermarking and extracting method based on SIFT feature |
CN109816584A (en) * | 2019-01-25 | 2019-05-28 | 燕山大学 | A kind of colour zero watermarking building method and extracting method |
CN111368585A (en) * | 2018-12-25 | 2020-07-03 | 中国科学院长春光学精密机械与物理研究所 | Weak and small target detection method, detection system, storage device and terminal equipment |
CN113129198A (en) * | 2021-04-29 | 2021-07-16 | 南京师范大学 | Zero watermark generation method and system and copyright infringement comparison method and system |
CN113658030A (en) * | 2021-08-18 | 2021-11-16 | 辽宁工程技术大学 | Low false alarm zero watermark algorithm based on regional XOR |
CN113781284A (en) * | 2021-06-30 | 2021-12-10 | 华南农业大学 | Zero watermark construction method based on depth attention self-encoder |
CN115311119A (en) * | 2022-10-09 | 2022-11-08 | 中国民航大学 | Three-dimensional image zero watermark embedding and extracting method capable of resisting geometric attack |
CN118469792A (en) * | 2024-07-11 | 2024-08-09 | 齐鲁工业大学(山东省科学院) | High-imperceptibility color watermark attack method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101504850A (en) * | 2009-03-12 | 2009-08-12 | 宁波大学 | Digital watermark treatment method for copyright protection |
CN102044055A (en) * | 2010-09-09 | 2011-05-04 | 中国矿业大学 | High-capacity digital watermarking method combining multi-scale characteristics and Chinese remainder theorem |
CN103955878A (en) * | 2014-04-11 | 2014-07-30 | 杭州电子科技大学 | DWT-SVD geometric attack prevention blind watermark method based on SIFT |
US20170091528A1 (en) * | 2014-03-17 | 2017-03-30 | Carnegie Mellon University | Methods and Systems for Disease Classification |
-
2017
- 2017-08-22 CN CN201710722220.9A patent/CN107507122B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101504850A (en) * | 2009-03-12 | 2009-08-12 | 宁波大学 | Digital watermark treatment method for copyright protection |
CN102044055A (en) * | 2010-09-09 | 2011-05-04 | 中国矿业大学 | High-capacity digital watermarking method combining multi-scale characteristics and Chinese remainder theorem |
US20170091528A1 (en) * | 2014-03-17 | 2017-03-30 | Carnegie Mellon University | Methods and Systems for Disease Classification |
CN103955878A (en) * | 2014-04-11 | 2014-07-30 | 杭州电子科技大学 | DWT-SVD geometric attack prevention blind watermark method based on SIFT |
Non-Patent Citations (2)
Title |
---|
JIANZHONG LI等: "A robust image watermarking scheme with kinoform in hybrid NSCT and SVD domain", 《2014 IEEE 7TH JOINT INTERNATIONAL INFORMATION TECHNOLOGY AND ARTIFICIAL INTELLIGENCE CONFERENCE》 * |
李琛等: "基于视觉感知的立体图像零水印算法研究", 《中国传媒大学学报自然科学版》 * |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109410115A (en) * | 2018-10-31 | 2019-03-01 | 山东省计算中心(国家超级计算济南中心) | The insertion of adaptive capacity image blind watermarking and extracting method based on SIFT feature |
CN111368585A (en) * | 2018-12-25 | 2020-07-03 | 中国科学院长春光学精密机械与物理研究所 | Weak and small target detection method, detection system, storage device and terminal equipment |
CN111368585B (en) * | 2018-12-25 | 2023-04-21 | 中国科学院长春光学精密机械与物理研究所 | Weak and small target detection method, detection system, storage device and terminal equipment |
CN109816584A (en) * | 2019-01-25 | 2019-05-28 | 燕山大学 | A kind of colour zero watermarking building method and extracting method |
CN113129198A (en) * | 2021-04-29 | 2021-07-16 | 南京师范大学 | Zero watermark generation method and system and copyright infringement comparison method and system |
CN113129198B (en) * | 2021-04-29 | 2024-01-12 | 南京师范大学 | Zero watermark generation method and system and copyright infringement comparison method and system |
CN113781284A (en) * | 2021-06-30 | 2021-12-10 | 华南农业大学 | Zero watermark construction method based on depth attention self-encoder |
CN113658030A (en) * | 2021-08-18 | 2021-11-16 | 辽宁工程技术大学 | Low false alarm zero watermark algorithm based on regional XOR |
CN115311119A (en) * | 2022-10-09 | 2022-11-08 | 中国民航大学 | Three-dimensional image zero watermark embedding and extracting method capable of resisting geometric attack |
CN115311119B (en) * | 2022-10-09 | 2022-12-23 | 中国民航大学 | Three-dimensional image zero watermark embedding and extracting method capable of resisting geometric attack |
CN118469792A (en) * | 2024-07-11 | 2024-08-09 | 齐鲁工业大学(山东省科学院) | High-imperceptibility color watermark attack method |
Also Published As
Publication number | Publication date |
---|---|
CN107507122B (en) | 2020-10-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107507122A (en) | Stereo-picture Zero watermarking method based on NSCT and SIFT | |
Wang et al. | Ternary radial harmonic Fourier moments based robust stereo image zero-watermarking algorithm | |
US9805296B2 (en) | Method and apparatus for decoding or generating multi-layer color QR code, method for recommending setting parameters in generation of multi-layer QR code, and product comprising multi-layer color QR code | |
CN103955878B (en) | DWT SVD resist geometric attackses blind watermark methods based on SIFT | |
Zhang et al. | Affine Legendre moment invariants for image watermarking robust to geometric distortions | |
CN103955880B (en) | DWT SVD Robust Blind Watermarking Scheme methods based on Zernike squares | |
CN102682418B (en) | Method for embedding and extracting multiple zero watermarks of digital image | |
CN109478243B (en) | Method for enhanced authentication of a material body | |
CN109034333A (en) | QR code dual anti-counterfeiting method based on rsa encryption and digital watermarking | |
CN104053074B (en) | Video watermarking method based on depth image and Otsu segmentation | |
CN104751402B (en) | A kind of method of robust reversible watermark insertion | |
CN110458747A (en) | The reversible watermarking algorithm of robust based on invariant features | |
CN102693522A (en) | Method for detecting region duplication and forgery of color image | |
CN104217387B (en) | A kind of image watermark based on qualified insertion is embedded, the method and apparatus of extraction | |
CN111861846A (en) | Electronic document digital watermark processing method and system | |
CN105787863A (en) | Linear canonical wavelet transform (LCWT)-based digital watermarking method | |
CN107067363A (en) | A kind of non-blind robust Spatial Domain Image Watermark method constant based on pixel order relation | |
Gul | A blind robust color image watermarking method based on discrete wavelet transform and discrete cosine transform using grayscale watermark image | |
Inamdar et al. | Offline handwritten signature based blind biometric watermarking and authentication technique using biorthogonal wavelet transform | |
CN104504643A (en) | Robustness digital water mark embedding and detection method based on local content features | |
CN106408503A (en) | Immune and digital image watermarking method for leveled rights management of digital works | |
CN111680549B (en) | Paper grain identification method | |
Escalante-Ramírez et al. | A perceptive approach to digital image watermarking using a brightness model and the hermite transform | |
CN102737380A (en) | Stereo image quality objective evaluation method based on gradient structure tensor | |
CN110428355A (en) | A kind of Watermarking for Color Images method based on compressed sensing and NSCT |
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 |