CN106023055B - Fragile reversible water mark method based on image local area - Google Patents
Fragile reversible water mark method based on image local area Download PDFInfo
- Publication number
- CN106023055B CN106023055B CN201610333977.4A CN201610333977A CN106023055B CN 106023055 B CN106023055 B CN 106023055B CN 201610333977 A CN201610333977 A CN 201610333977A CN 106023055 B CN106023055 B CN 106023055B
- Authority
- CN
- China
- Prior art keywords
- sequence
- characteristic
- detected
- circle
- image
- 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.)
- Active
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
- G06T1/0042—Fragile watermarking, e.g. so as to detect tampering
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Editing Of Facsimile Originals (AREA)
Abstract
A kind of fragile reversible water mark method based on image local area, specific steps include: the sampling of (1) characteristic image;(2) initial characteristics point sequence is extracted;(3) extension feature point sequence is calculated;(4) scale feature point sequence is obtained;(5) non-characteristic circle sequence of crossing the border is obtained;(6) non-overlap characteristic circle sequence is obtained;(7) it is embedded in watermark;(8) characteristic image samples;(9) initial characteristics point sequence is extracted;(10) extension feature point sequence is calculated;(11) scale feature point sequence is obtained;(12) non-characteristic circle sequence of crossing the border is obtained;(13) non-overlap characteristic circle sequence is obtained;(14) watermark is extracted.The present invention determines area-of-interest using Harris-Laplace operator selected characteristic point, adjusts radius scale factor and determines area-of-interest size, improve watermark capacity and visual quality, has area-of-interest tampering detection ability.
Description
Technical field
The invention belongs to digital information safety technical fields, further relate to digital figure watermark insertion and extractive technique
Fragile reversible water mark method of one of the field based on image local area.The present invention can be used for digital picture in network environment
Tampering detection, realize the content authentication of digital picture, for copyright protection, infringement retrospect important foundation is provided.
Background technique
Identification information is embedded into the technology in information carrier as a kind of by the current information age, and watermark is successfully answered
For in the information carriers such as digital picture, audio, video and network communication, playing the role of copyright protection, infringement retrospect.But
It is that often for good and all modification information carrier information is difficult to meet conventional watermark technology although this modification is very little
Military, medical treatment and the undistorted requirement of judicial domain information carrier.For this problem, reversible water mark technology is come into being, it
By restoring watermark and host image without distortions, the integrality of watermark and host image ensure that.It is all due to reversible water mark
More advantages are widely studied and are paid close attention at present, and achieve some breakthroughs.However, existing can be against the current
Most of impression method is designed based on entire image, fail to consider well the importance of interesting image regions with it is special
Property.Such as in medical image, doctor is often only concerned organ or diseased region in medical image.Therefore, how to design effectively
Reversible water mark method for area-of-interest content protecting it is most important.
Cui get Long, left patent " a kind of region-of-interest authentication and tampering detection digital watermark method " (Shen for respecting dragon application
Please number: 201210443250.3, application publication number: CN 102945542A) disclose a kind of region-of-interest authentication and distort inspection
Survey digital watermark method.Area-of-interest is defined by the user in watermark built-in end in this method, is generating watermark information and Hash code
Watermark information is embedded into the coefficient of wavelet decomposition of image background regions later.In receiving end, by comparing area-of-interest
Hash code and watermark reconstruction carry out double authentication.Shortcoming existing for this method is: due to area-of-interest in this method
It needs to be determined by user oneself, the lager time cost of consumption;And the otherness meeting that different user selects area-of-interest
The capacity and the visual quality containing watermarking images for influencing watermark.
Q.Gu, T.Gao are in paper " A novel reversible watermarking scheme based on
block energy difference for medical images”(《2012Joint 6th International
Conference on Soft Computing and Intelligent Systems(SCIS)and 13th
International Symposium on Advanced Intelligent Systems (ISIS) ", 2012,232-237)
In propose a kind of reversible water mark method based on medical image capacity volume variance.This method is not overlapped original image point
Block carries out integer wavelet transform to every sub-image respectively, is embedded watermark data into according to the capacity volume variance of each image subblock whole
In the low frequency sub-band of type wavelet transformation.Shortcoming existing for this method is: due to this method be in regions of non-interest it is embedding
Enter watermark, cannot achieve effective detection of watermark after area-of-interest is tampered, do not have in this way interested
Region tampering detection ability.
Summary of the invention
It is a kind of based on image local area it is an object of the invention to aiming at the shortcomings of the prior art, propose
Fragile reversible water mark method not only increases the embedding capacity of watermark and the visual quality containing watermarking images, and has and feel emerging
Interesting region tampering detection ability.
Realizing the concrete thought of the object of the invention is, in watermark telescopiny: carrying out characteristic pattern to carrier image first
As sampling, then according to Harris-Laplace operator extraction initial characteristics point sequence, extension feature point sequence is then successively obtained
Column, scale feature point sequence, non-cross the border characteristic circle sequence and non-overlap characteristic circle sequence, finally successively in non-overlap characteristic circle sequence
Each non-overlap characteristic circle in column carries out watermark insertion, obtains containing watermarking images;In watermark extraction process: first to be checked
Altimetric image carries out characteristic image sampling and connects then according to Harris-Laplace operator extraction initial characteristics point sequence to be detected
Successively obtain extension feature point sequence to be detected, scale feature point sequence to be detected, it is to be detected it is non-cross the border characteristic circle sequence with
Non-overlap characteristic circle sequence to be detected, finally successively each non-overlap characteristic circle in non-overlap characteristic circle sequence to be detected into
Row watermark extracting, image and watermark after being restored.
The present invention includes two processes of watermark insertion and watermark extracting;
The specific steps of watermark telescopiny of the present invention are as follows:
(1) characteristic image samples:
Carrier image is extracted one piece of square matrix of V × V pixel by (1a) from upper left side, this square matrix is divided into
Size is 2 × 2 pixels and the sub-block that does not overlap, sorts left to right to obtain carrier image block sequence;
(1b) cannot divided remainder after abandoning piecemeal;
(1c) successively chooses each sub-block in carrier image block sequence, calculates sub-block in selected carrier image block sequence
Average pixel value obtains mean pixel value sequence A;
Mean pixel value sequence A is converted characteristic image H by (1d);
(2) according to the following formula, initial characteristics point is extracted from characteristic image, obtains initial characteristics point sequence:
F=harrisLaplace (H)
Wherein, F indicates that initial characteristics point sequence, harrisLaplace indicate to carry out the calculating of Harris-Laplace operator
Operation, H indicate characteristic image;
(3) extension feature point sequence is calculated:
Each characteristic point in initial characteristics point sequence is successively chosen, by the characteristic point in selected initial characteristics point sequence
Corresponding line position, which is set, expands as original twice with column position, and be expanded characteristic point sequence;
(4) scale feature point sequence is obtained:
(4a) successively chooses each characteristic point in extension feature point sequence, calculates special in selected extension feature point sequence
The corresponding scale of sign point, obtains characteristic point scaling sequence;
(4b) is in scale threshold interval [T1,T2] in, choose all scales met in characteristic point scaling sequence, wherein 1
≤T1< T2≤16;
(4c) successively using the corresponding extension feature point of the scale in characteristic point scaling sequence as scale feature point, obtains ruler
Spend characteristic point sequence;
(5) non-characteristic circle sequence of crossing the border is obtained:
(5a) chooses a scale feature point from scale feature point sequence, is with the position where selected characteristic point
The center of circle is justified by radius work of the product of the scale of characteristic point and given radius scale factor, using the circle as non-feature of crossing the border
Circle;
(5b) judges whether non-characteristic circle of crossing the border is more than the boundary of carrier image, if so, give up the non-characteristic circle of crossing the border,
Otherwise, step (5c) is executed;
(5c) judges whether to have chosen all scale feature points in scale feature point sequence, if so, step (5d) is executed,
It is no to then follow the steps (5a);
All non-characteristic circles of crossing the border are formed non-characteristic circle sequence of crossing the border by (5d);
(6) non-overlap characteristic circle sequence is obtained:
(6a) chooses a non-characteristic circle of crossing the border from non-characteristic circle sequence of crossing the border, and selected non-characteristic circle of crossing the border is put
Enter in non-overlap characteristic circle sequence;
(6b) chooses a non-characteristic circle of crossing the border from non-characteristic circle sequence of crossing the border, and judges selected non-characteristic circle of crossing the border
Whether overlap on disc with non-overlap characteristic circle sequence, if so, otherwise giving up the non-characteristic circle of crossing the border executes step
Suddenly (6c);
Non- characteristic circle of crossing the border is put into non-overlap characteristic circle sequence by (6c);
(6d) judges whether to have chosen all non-characteristic circles of crossing the border in non-characteristic circle sequence of crossing the border, if so, executing step
(7), no to then follow the steps (6b);
(7) it is embedded in watermark:
(7a) chooses a non-overlap characteristic circle from non-overlap characteristic circle sequence, calculates selected non-overlap characteristic circle
Inscribed square;
(7b) according to the following formula, is embedded in watermark in the inscribed square of selection, obtains the inscribed square containing watermark:
G=embedWM (Q, W)
Wherein, G indicates that the inscribed square containing watermark, embedWM indicate that watermark embedding operation, Q indicate non-overlap feature
Round inscribed square, W indicate the watermark of quasi- insertion;
(7c) judges whether to have chosen all non-characteristic circles of crossing the border in non-overlap characteristic circle sequence, if so, obtaining aqueous
Otherwise watermark image executes step (7a);
The specific steps of the watermark extraction process are as follows:
(8) characteristic image sampling to be detected:
Image to be detected is extracted one piece of V ' × V ' pixel square matrix from upper left side by (8a), by this square matrix
It is divided into the sub-block that size is 2 × 2 pixels and does not overlap, sorts left to right to obtain image to be detected block sequence;
(8b) cannot divided remainder after abandoning piecemeal;
(8c) successively chooses each sub-block in image to be detected block sequence, calculates selected image to be detected block sequence neutron
The average pixel value of block obtains mean pixel value sequence A ' to be detected;
Mean pixel value sequence A ' to be detected is converted characteristic image H ' to be detected by (8d);
(9) according to the following formula, initial characteristics point to be detected is extracted from characteristic image to be detected, obtains initial characteristics to be detected
Point sequence:
F '=harrisLaplace (H ')
Wherein, F ' expression initial characteristics point sequence to be detected, harrisLaplace indicate to carry out Harris-Laplace calculation
Sub- calculating operation, H ' expression characteristic image to be detected;
(10) extension feature point sequence to be detected is calculated:
Each characteristic point in initial characteristics point sequence to be detected is successively chosen, by selected initial characteristics point sequence to be detected
The corresponding line position of characteristic point in column, which is set, expands as original twice with column position, obtains extension feature point sequence to be detected;
(11) scale feature point sequence to be detected is obtained:
(11a) successively chooses each characteristic point in extension feature point sequence to be detected, calculates selected extension to be detected
The corresponding scale of characteristic point in characteristic point sequence obtains characteristic point scaling sequence to be detected;
(11b) is in scale threshold interval [T1,T2] in, all scales met in characteristic point scaling sequence to be detected are chosen,
Wherein, 1≤T1< T2≤16;
(11c) is successively using the corresponding extension feature point to be detected of the scale in characteristic point scaling sequence to be detected as scale
Characteristic point obtains scale feature point sequence to be detected;
(12) non-characteristic circle sequence of crossing the border to be detected is obtained:
(12a) chooses a scale feature point from scale feature point sequence to be detected, where selected characteristic point
Position be the center of circle, be that radius is made to justify using the product of the scale of characteristic point and given radius scale factor, using the circle as non-
It crosses the border characteristic circle;
(12b) judges whether non-characteristic circle of crossing the border is more than the boundary of image to be detected, if so, giving up the non-feature of crossing the border
Otherwise circle executes step (12c);
(12c) judges whether to have chosen all scale feature points in scale feature point sequence to be detected, if so, executing step
Suddenly (12d), it is no to then follow the steps (12a);
All non-characteristic circles of crossing the border are formed non-characteristic circle sequence of crossing the border to be detected by (12d);
(13) non-overlap characteristic circle sequence to be detected is obtained:
(13a) chooses a non-characteristic circle of crossing the border from non-characteristic circle sequence of crossing the border to be detected, non-crosses the border selected
Characteristic circle is put into non-overlap characteristic circle sequence to be detected;
(13b) chooses a non-characteristic circle of crossing the border from non-characteristic circle sequence of crossing the border to be detected, judge it is selected it is non-more
Whether boundary's characteristic circle overlaps on disc with non-overlap characteristic circle sequence to be detected, if so, giving up the non-feature of crossing the border
Otherwise circle executes step (13c);
Non- characteristic circle of crossing the border is put into non-overlap characteristic circle sequence to be detected by (13c);
(13d) judges whether to have chosen all non-characteristic circles of crossing the border in non-characteristic circle sequence of crossing the border to be detected, if so, holding
Row step (14), it is no to then follow the steps (13b);
(14) watermark is extracted:
(14a) chooses a non-overlap characteristic circle from non-overlap characteristic circle sequence to be detected, calculates selected non-heavy
The inscribed square of folded characteristic circle;
(14b) according to the following formula, extracts watermark in the inscribed square of selection, obtains the inscribed square containing watermark:
(G ', W ')=extractWM (Q ')
Wherein, the inscribed square after G ' expression reduction, the watermark after W ' expression reduction, extractWM indicate that watermark mentions
Extract operation, the inscribed square of Q ' expression non-overlap characteristic circle;
(14c) judges whether to have chosen all non-overlap characteristic circles in non-overlap characteristic circle sequence to be detected, if so,
Otherwise image after to reduction executes step (14a).
Compared with prior art, the invention has the following advantages that
First, since the present invention is during watermark is embedded in and is extracted, selected by using Harris-Laplace operator
Take initial characteristics points and determine area-of-interest by screening, overcome area-of-interest in the prior art must be by with
Family oneself is specified, the larger disadvantage of time loss cost, so that there is the present invention area-of-interest to choose automatically, time loss is small
The advantages of.
Second, since the present invention is in watermark insertion and extraction process, by being embedded in watermark in the region of interest, overcome
Some prior arts cannot achieve the effective detection of watermark after area-of-interest is tampered, so that the present invention has sense
The advantages of interest region tampering detection.
Third determines that sense is emerging by adjusting given radius scale factor since the present invention is in watermark telescopiny
Interesting area size, overcomes some watermark capacities in the prior art and the shortcomings that visual quality can not be adjusted effectively, so that this hair
It is bright to have the advantages that watermark capacity and visual quality are good.
Detailed description of the invention
Fig. 1 is watermark insertion flow diagram of the invention;
Fig. 2 is watermark extracting flow diagram of the invention;
Fig. 3 is experiment test image of the invention.
Specific embodiment
The present invention is described in detail with reference to the accompanying drawing.
Referring to attached drawing 1, watermark Embedded step of the invention is as follows.
Step 1, characteristic image samples.
Carrier image I is extracted one piece of square of V × V pixel by the carrier image I that given size is M × N from upper left side
Matrix, wherein V=2 × c, c are to meetUnder the conditions of maximum positive integer,It indicates to be rounded downwards
Operation, min expression are minimized operation, and M and N respectively indicate the line number and columns of carrier image.
This square matrix is divided into the sub-block that size is 2 × 2 pixels and does not overlap, sorts left to right to obtain carrier image
Block sequence B={ B1,...,Bφ,...Bc×c, wherein BφThe φ sub-block in expression carrier image block sequence B, 1≤φ≤(c ×
It c), cannot divided remainder after discarding piecemeal.
Each sub-block in carrier image block sequence is successively chosen, according to the following formula, is calculated in selected carrier image block sequence
The average pixel value of sub-block obtains mean pixel value sequence A:
Ai=(Bi1+Bi2+Bi3+Bi4)/4
Wherein, AiIndicate the average pixel value of i-th of sub-block in carrier image block sequence, Bi1Indicate carrier image block sequence
In be located at pixel value at upper left position, B in i-th of sub-blocki2It indicates to be located at a left side in carrier image block sequence in i-th of sub-block
The pixel value of lower angular position, Bi3Indicate the pixel value for being located at upper right angular position in carrier image block sequence in i-th of sub-block,
Bi4It indicates to be located at the pixel value at lower right position in carrier image block sequence in i-th of sub-block.
According to the following formula, characteristic image H is converted by mean pixel value sequence A:
H (q, t)=Ak
Wherein, H (q, t) indicates the pixel value of q row t column in characteristic image H, AkIt indicates in mean pixel value sequence A
K-th of value, k=q × c+t-c, c indicate meetUnder the conditions of maximum positive integer,Indicate to
Lower floor operation, min expression are minimized operation, and M and N respectively indicate the line number and columns of carrier image, 1≤q≤c, 1≤t
≤c。
Step 2, initial characteristics point sequence is extracted.
According to document " D.Lowe.Object Recognition from Local Scale-Invariant
The feature proposed in Features.Computer Science Department, vol.2, pp.1150-1157, Sep 1999 "
Point extracting method extracts characteristic point using Harris-Laplace operator from characteristic image, obtains initial characteristics according to the following formula
Point sequence F:
F=harrisLaplace (H)
Wherein, F indicates that initial characteristics point sequence, harrisLaplace indicate to carry out the calculating of Harris-Laplace operator
Operation, H indicate characteristic image.
Step 3, extension feature point sequence is calculated.
Each characteristic point in initial characteristics point sequence F is successively chosen, by the feature in selected initial characteristics point sequence F
The corresponding line position of point, which is set, expands as original twice with column position, the characteristic point sequence that is expanded U.
Step 4, scale feature point sequence is obtained.
Each characteristic point in extension feature point sequence U is successively chosen, according to the following formula, calculates selected extension feature point sequence
The corresponding scale of characteristic point in U is arranged, characteristic point scaling sequence S is obtained:
Sj=getScale (Uj)
Wherein, SjIndicate that j-th of scale in characteristic point scaling sequence S, getScale indicate dimension calculation operation, UjIt indicates
J-th of extension feature point in extension feature point sequence.
In scale threshold interval [T1,T2] in, choose all scales met in characteristic point scaling sequence S, wherein T1With
T2The respectively minimum value and maximum value of scale threshold interval, T1< T2。
Successively using the corresponding extension feature point of characteristic point scaling sequence S mesoscale as scale feature point, scale spy is obtained
Levy point sequence C.
Step 5, non-characteristic circle sequence of crossing the border is obtained.
Step 1 chooses a scale feature point, from scale feature point sequence C with the position where selected characteristic point
It is set to the center of circle, with the scale S of characteristic pointηProduct with given radius scale factor λ is that radius work is justified, and is got over using the circle as non-
Boundary characteristic circle Zη, wherein SηIndicate the η scale in characteristic point scaling sequence S, ZηIndicate that the η scale feature point is corresponding non-
It crosses the border characteristic circle.
Step 2, judges whether the non-characteristic circle Z that crosses the border is more than the boundary of carrier image I, if so, giving up the non-spy that crosses the border
Otherwise sign circle executes the step 3 in step 5.
Step 3 judges whether to have chosen all scale feature points in scale feature point sequence C, if so, executing step 5
In step 4, otherwise, execute step 5 in step 1.
All non-characteristic circles of crossing the border are formed the non-characteristic circle sequence Z={ Z that crosses the border by step 41,...,Zθ,...,Zσ,
In, ZθIndicate that θ non-characteristic circles of crossing the border, σ indicate the number of non-characteristic circle of crossing the border in non-characteristic circle sequence of crossing the border.
Step 6, non-overlap characteristic circle sequence is obtained.
Step 1 successively chooses the corresponding characteristic point of each characteristic circle in the non-characteristic circle sequence Z that crosses the border, and according to the following formula, calculates
The intensity of the corresponding characteristic point of characteristic circle in the selected characteristic circle sequence Z that crosses the border, obtains characteristic point sequence of intensity P:
Pτ=getPower (Lτ)
Wherein, PτIndicate the τ characteristic point intensity in characteristic point sequence of intensity P, getPower indicates characteristic point intensitometer
Calculate operation, LτIndicate the corresponding characteristic point of the τ characteristic circle in the non-characteristic circle sequence Z that crosses the border.
Step 2, according to the non-corresponding characteristic point sequence of intensity P of characteristic circle sequence Z that crosses the border, by the non-characteristic circle sequence Z that crosses the border
In each non-characteristic circle of crossing the border be ranked up from big to small according to its corresponding characteristic point intensity.
Step 3 chooses first non-characteristic circle of crossing the border, by the selected non-spy that crosses the border from the non-characteristic circle sequence Z that crosses the border
Sign circle is put into non-overlap characteristic circle sequence O.
Step 4 successively chooses a non-characteristic circle of crossing the border from the non-characteristic circle sequence Z that crosses the border, judge it is selected it is non-more
Whether boundary's characteristic circle overlaps on disc with each non-overlap characteristic circle in non-overlap characteristic circle sequence O, if so, house
The non-characteristic circle of crossing the border is abandoned, otherwise, executes the step 5 in step 6.
Non- characteristic circle of crossing the border is put into non-overlap characteristic circle sequence O by step 5.
Step 6 judges whether to have chosen all non-characteristic circles of crossing the border in the non-characteristic circle sequence Z that crosses the border, if so, executing step
Rapid 7, otherwise, execute the step 4 in step 6.
Step 7, it is embedded in watermark.
Step 1 chooses a non-overlap characteristic circle from non-overlap characteristic circle sequence O, according to the following formula, calculates selected
The inscribed square of non-overlap characteristic circle:
Wherein, L1Indicate that the inscribed square upper left angle point of non-overlap characteristic circle is corresponding in the plane in carrier image institute
Abscissa value, L2Indicate the inscribed square upper left angle point of non-overlap characteristic circle in carrier image institute vertical seat corresponding in the plane
Scale value, D1Indicate the inscribed square bottom right angle point of non-overlap characteristic circle in carrier image institute abscissa corresponding in the plane
Value, D2Indicate the inscribed square bottom right angle point of non-overlap characteristic circle in carrier image institute ordinate value corresponding in the plane, x
The center of circle of non-overlap characteristic circle is indicated in carrier image institute abscissa value corresponding in the plane, y indicates non-overlap characteristic circle
The center of circle indicates the side of the inscribed square of non-overlap characteristic circle in carrier image institute ordinate value corresponding in the plane, E,Wherein,Indicate that downward floor operation, R indicate the radius of selected non-overlap characteristic circle.
Step 2 utilizes document " X.Gao, L.An, Y.Yuan, D.Tao, X.Li.Lossless Data Embedding
Using Generalized Statistical Quantity Histogram.IEEE Transactions on
It is proposed in Circuits and Systems for Video Technology, vol.21, pp.1061-1070, Aug 2011 "
Watermark embedding method, according to the following formula, selection it is inscribed square in be embedded in watermark, obtain the inscribed square containing watermark:
G=embedWM (Q, W)
Wherein, G indicates that the inscribed square containing watermark, embedWM indicate that watermark embedding operation, Q indicate non-overlap feature
Round inscribed square, W indicate the watermark of quasi- insertion.
Step 3 judges whether to have chosen all non-characteristic circles of crossing the border in non-overlap characteristic circle sequence Z, if so, being contained
Watermarking images IW, otherwise, execute the step 1 in step 7.
1 watermark insertion may be implemented to step 7 through the above steps, obtains I containing watermarking imagesW。
Referring to attached drawing 2, watermark extraction step of the invention is as follows.
Step 8, characteristic image sampling to be detected.
Given size is M ' × N ' image to be detected I ', and image to be detected I ' is extracted V ' × V ' pixel from upper left side
One piece of square matrix to be detected, wherein V '=2 × d, d are to meetUnder the conditions of maximum it is just whole
Number,Indicate downward floor operation, min expression is minimized operation, and M ' and N ' respectively indicate the line number and column of image to be detected
Number.
This square matrix to be detected is divided into the sub-block that size is 2 × 2 pixels and does not overlap, sort left to right to obtain to
Detection image block sequence B '={ B '1,...,B′γ,...B′d×d, wherein BγIndicate the γ son in image to be detected block sequence B
Block, 1≤γ≤(d × d), abandon piecemeal after cannot divided remainder.
Each sub-block in image to be detected block sequence is successively chosen, according to the following formula, calculates selected image to be detected block sequence
The average pixel value of sub-block in column obtains mean pixel value sequence A ' to be detected:
A′α=(B 'α1+B′α2+B′α3+B′α4)/4
Wherein, A 'αIndicate the average pixel value of the α sub-block in image to be detected block sequence, B 'α1Indicate image to be detected
The pixel value being located in the α sub-block at upper left position in block sequence, B 'α2Indicate the α son in image to be detected block sequence
It is located at the pixel value of lower-left angular position, B ' in blockα3It indicates to be located at upper right corner position in image to be detected block sequence in the α sub-block
Set the pixel value at place, B 'α4It indicates to be located at the pixel value at lower right position in image to be detected block sequence in the α sub-block.
According to the following formula, characteristic image to be detected is converted by mean pixel value sequence A ' to be detected:
H ' (μ, υ)=A 'β
Wherein, H ' (μ, υ) indicates the pixel value of μ row υ column in characteristic image H ' to be detected, A 'βIndicate to be detected flat
The β value in equal sequence of pixel values A ', β=μ × d+ υ-d, d indicate satisfactionUnder the conditions of maximum
Positive integer,Indicate downward floor operation, min expression is minimized operation, and M ' and N ' respectively indicate the line number of image to be detected
With columns, 1≤μ≤d, 1≤υ≤d.
Step 9, initial characteristics point sequence to be detected is extracted.
According to document " D.Lowe.Object Recognition from Local Scale-Invariant
The feature proposed in Features.Computer Science Department, vol.2, pp.1150-1157, Sep 1999 "
Point extracting method extracts characteristic point from characteristic image to be detected using Harris-Laplace operator according to the following formula, obtain to
Detect initial characteristics point sequence F ':
F '=harrisLaplace (H ')
Wherein, F ' expression initial characteristics point sequence to be detected, harrisLaplace indicate to carry out Harris-Laplace calculation
Sub- calculating operation, H ' expression characteristic image.
Step 10, extension feature point sequence to be detected is calculated.
Each characteristic point in initial characteristics point sequence F ' to be detected is successively chosen, by selected initial characteristics point to be detected
The corresponding line position of characteristic point in sequence F ', which is set, expands as original twice with column position, obtains extension feature point sequence to be detected
U′。
Step 11, scale feature point sequence to be detected is obtained.
Each characteristic point in extension feature point sequence U ' to be detected is successively chosen, according to the following formula, is calculated selected to be checked
The corresponding scale of the middle characteristic point of extension feature point sequence U ' is surveyed, characteristic point scaling sequence S ' to be detected is obtained:
S′ε=getScale (U 'ε)
Wherein, S 'εIndicate that the ε feature point scale in characteristic point scaling sequence to be detected, getScale indicate scale meter
Calculate operation, U 'εIndicate the ε extension feature point in extension feature point sequence to be detected.
In scale threshold interval [T1,T2] in, all scales met in characteristic point scaling sequence S ' to be detected are chosen,
In, T1And T2The respectively minimum value and maximum value of scale threshold interval, T1< T2。
Successively using the corresponding extension feature point of the scale in characteristic point scaling sequence S ' to be detected as scale feature point, obtain
To scale feature point sequence C ' to be detected.
Step 12, non-characteristic circle sequence of crossing the border to be detected is obtained.
Step 1, from scale feature point sequence C ' to be detected one scale feature point of middle selection, with selected characteristic point institute
Position be the center of circle, with the scale S ' of characteristic pointψProduct with given radius scale factor λ is that radius work is justified, by the circle
As the non-characteristic circle Z ' that crosses the border to be detectedψ, wherein S 'ψIndicate the ψ scale in characteristic point scaling sequence S ' to be detected, Z 'ψTable
Show the corresponding non-characteristic circle of crossing the border of the ψ scale feature point.
Step 2 judges the non-characteristic circle Z ' that crosses the borderψWhether it is more than the boundary of image to be detected I ', non-crosses the border if so, giving up this
Otherwise characteristic circle executes the step 3 in step 12.
Step 3 judges whether to have chosen all scale feature points in scale feature point sequence C ' to be detected, if so, holding
Row step 4, the no step 1 thened follow the steps in 12.
All non-characteristic circles of crossing the border are formed non-characteristic circle sequence of crossing the border to be detected by step 4Wherein, Z 'φIndicate φ non-features of crossing the border in non-characteristic circle sequence of crossing the border to be detected
Circle,Indicate the number of non-characteristic circle of crossing the border to be detected.
Step 13, non-overlap characteristic circle sequence to be detected is obtained.
Step 1 successively chooses the corresponding characteristic point of each characteristic circle in the non-characteristic circle sequence Z ' that crosses the border to be detected, under
Formula calculates the intensity of the selected middle characteristic circle of characteristic circle sequence Z ' corresponding characteristic point to be detected of crossing the border, obtains spy to be detected
Sign point sequence of intensity P ':
P′ξ=getPower (L 'ξ)
Wherein, P 'ξIndicate the ξ characteristic point intensity in characteristic point sequence of intensity P ', getPower indicates characteristic point intensity
Calculating operation, L 'ξIndicate the corresponding characteristic point of the ξ characteristic circle in the non-characteristic circle sequence Z ' that crosses the border to be detected.
Step 2, according to the non-corresponding characteristic point sequence of intensity P ' of characteristic circle sequence Z ' that crosses the border to be detected, by it is to be detected it is non-more
Each non-characteristic circle of crossing the border is ranked up from big to small according to its corresponding characteristic point intensity in boundary characteristic circle sequence Z '.
Step 3 will be selected non-from first non-characteristic circle of crossing the border of the non-middle selection of the characteristic circle sequence Z ' that crosses the border to be detected
Characteristic circle of crossing the border is put into non-overlap characteristic circle sequence O ' to be detected.
Step 4, successively chooses a non-characteristic circle of crossing the border from the non-characteristic circle sequence Z ' that crosses the border to be detected, selected by judgement
Non- characteristic circle of crossing the border whether occur on disc with each non-overlap characteristic circle in non-overlap characteristic circle sequence O ' to be detected
Otherwise overlapping, executes the step 5 in step 13 if so, giving up the non-characteristic circle of crossing the border.
Non- characteristic circle of crossing the border is put into non-overlap characteristic circle sequence O ' to be detected by step 5.
Step 6 judges whether to have chosen all non-characteristic circles of crossing the border in the non-characteristic circle sequence Z ' that crosses the border to be detected, if
It is to execute step 14, the no step 4 thened follow the steps in 13.
Step 14, watermark is extracted.
Step 1 according to the following formula, is calculated from non-overlap characteristic circle sequence O ' to be detected one non-overlap characteristic circle of middle selection
The inscribed square of selected non-overlap characteristic circle:
Wherein, L '1Indicate that the inscribed square upper left angle point of non-overlap characteristic circle is right in the plane in image to be detected institute
The abscissa value answered, L '2Indicate that the inscribed square upper left angle point of non-overlap characteristic circle is right in the plane in image to be detected institute
The ordinate value answered, D '1Indicate that the inscribed square bottom right angle point of non-overlap characteristic circle is right in the plane in image to be detected institute
The abscissa value answered, D '2Indicate that the inscribed square bottom right angle point of non-overlap characteristic circle is right in the plane in image to be detected institute
The ordinate value answered, the center of circle of x ' expression non-overlap characteristic circle is in image to be detected institute abscissa value corresponding in the plane, y '
The center of circle of non-overlap characteristic circle is indicated in carrier image institute to be detected ordinate value corresponding in the plane, E ' expression non-overlap is special
The side of the inscribed square of circle is levied,Wherein,Indicate downward floor operation, it is non-selected by R ' expression
The radius of overlapping feature circle.
Step 2 utilizes document " X.Gao, L.An, Y.Yuan, D.Tao, X.Li.Lossless Data Embedding
Using Generalized Statistical Quantity Histogram.IEEE Transactions on
It is proposed in Circuits and Systems for Video Technology, vol.21, pp.1061-1070, Aug 2011 "
Watermark embedding method, according to the following formula, selection it is inscribed square in be embedded in watermark, obtain the inscribed square containing watermark:
(G ', W ')=extractWM (Q ')
Wherein, the inscribed square after G ' expression reduction, the watermark after W ' expression reduction, extractWM indicate that watermark mentions
Extract operation, the inscribed square of Q ' expression non-overlap characteristic circle.
Step 3 judges whether to have chosen all non-characteristic circles of crossing the border in non-overlap characteristic circle sequence to be detected, if so,
Otherwise image after being restored executes the step 1 in step 14.
8 watermark insertion may be implemented to step 14 through the above steps, the water after image I ' and reduction after being restored
Print W '.
Effect of the present invention is further described below with reference to emulation experiment test image.
1. emulation experiment condition:
The software environment for realizing the method for the present invention is the MATLAB 2015a of U.S. Mathworks company exploitation, system fortune
Row environment is Linux.Referring to attached drawing 3, Fig. 3 (a) and Fig. 3 (b) are emulation experiment test image, and size is 512 × 512.Emulation
The performance of the method for the present invention is tested in experiment in terms of capacity, visual quality and tampering detection ability three.For simplicity, by
The method of the present invention is denoted as RROI, by Q.Gu, T.Gao in paper " A novel reversible watermarking scheme
The method proposed in based on block energy difference for medical images " is denoted as BED.
2. emulation experiment content:
Experiment 1: Capacity Simulation experiment:
Emulation experiment capacity in the present invention refers to the watermark digit that most multipotency is embedded in carrier image, respectively in test chart
Watermark embedding operation is carried out using both methods as in, obtains detecting the watermark capacity containing watermarking images containing watermarking images, it is single
Position is bit, obtains that the results are shown in Table 1.By 1 result of table as it can be seen that the method for the present invention has higher water than prior art BED
Print capacity.
1. watermark capacity of table
Experiment 2: visual quality emulation experiment:
The present invention uses both in test image respectively using objective indicator Y-PSNR PSNR as judging basis
Method carries out watermark embedding operation, obtains detecting the visual quality containing watermarking images containing watermarking images.PSNR is expressed as
Wherein, M × N is the size of carrier image I,I indicates carrier
Image, IWIt indicates to contain watermarking images, I (i, j) is the pixel value that carrier image is arranged in the i-th row jth, IW(i, j) is containing watermarking images
In the pixel value of the i-th row jth column, obtain that the results are shown in Table 2.
2. visual quality of table
By 2 result of table as it can be seen that the method for the present invention has higher visual quality than prior art BED.
Experiment 3: tampering detection emulation experiment:
Referring to attached drawing 3, Fig. 3 (c) is that Fig. 3 (a) distorts the image obtained after area-of-interest, and Fig. 3 (d) is that Fig. 3 (b) is distorted
The image obtained after area-of-interest.Determine that the watermark of both methods is usurped by comparing watermark similarity S in emulation experiment
Change detectability.S is expressed as
Wherein, sum indicates sum operation, and bitxor indicates xor operation, WαIndicate the watermark of insertion, WβIt indicates from distorting
The watermark extracted in image afterwards, M indicate the length of insertion watermark.When S is less than 1, illustrate that watermark can effectively judge image
It distorts, and when S is equal to 1, it since the watermark of extraction and the watermark of insertion are identical, cannot achieve tampering detection, obtain
The results are shown in Table 3.
3. watermark similarity of table
By 3 result of table as it can be seen that compared with prior art BED, what the method for the present invention can be realized area-of-interest distorts inspection
It surveys.
Claims (7)
1. a kind of fragile reversible water mark method based on image local area, including two processes of watermark insertion and watermark extracting;
The specific steps of the watermark telescopiny are as follows:
(1) characteristic image samples:
Carrier image is extracted one piece of square matrix of V × V pixel by (1a) from upper left side, this square matrix is divided into size
For 2 × 2 pixels and the sub-block that does not overlap, sort left to right to obtain carrier image block sequence;Wherein, V=2 × c, c are to meetUnder the conditions of maximum positive integer,Indicate downward floor operation, min expression is minimized operation, M
The line number and columns of carrier image are respectively indicated with N;
(1b) cannot divided remainder after abandoning piecemeal;
(1c) successively chooses each sub-block in carrier image block sequence, and sub-block is averaged in the selected carrier image block sequence of calculating
Pixel value obtains mean pixel value sequence A;
(1d) is converted into characteristic image H according to the following formula, by mean pixel value sequence A;
H (q, t)=Ak
Wherein, H (q, t) indicates the pixel value of q row t column in characteristic image H, AkIndicate the kth in mean pixel value sequence A
A value, k=q × c+t-c, c indicate to meetUnder the conditions of maximum positive integer,It indicates to be rounded downwards
Operation, min expression are minimized operation, and M and N respectively indicate the line number and columns of carrier image, 1≤q≤c, 1≤t≤c;
(2) according to the following formula, initial characteristics point is extracted from characteristic image, obtains initial characteristics point sequence:
F=harrisLaplace (H)
Wherein, F indicates that initial characteristics point sequence, harrisLaplace indicate to carry out Harris-Laplace operator calculating operation,
H indicates characteristic image;
(3) extension feature point sequence is calculated:
Each characteristic point in initial characteristics point sequence is successively chosen, the characteristic point in selected initial characteristics point sequence is corresponding
Line position set and expand as original twice with column position, be expanded characteristic point sequence;
(4) scale feature point sequence is obtained:
(4a) successively chooses each characteristic point in extension feature point sequence, calculates characteristic point in selected extension feature point sequence
Corresponding scale obtains characteristic point scaling sequence;
(4b) is in scale threshold interval [T1,T2] in, choose all scales met in characteristic point scaling sequence, wherein 1≤T1
< T2≤16;
(4c) successively using the corresponding extension feature point of the scale in characteristic point scaling sequence as scale feature point, obtains scale spy
Levy point sequence;
(5) non-characteristic circle sequence of crossing the border is obtained:
(5a) chooses a scale feature point from scale feature point sequence, is circle with the position where selected characteristic point
The heart is justified by radius work of the product of the scale of characteristic point and given radius scale factor, using the circle as non-characteristic circle of crossing the border;
(5b) judges whether non-characteristic circle of crossing the border is more than the boundary of carrier image, if so, give up the non-characteristic circle of crossing the border, it is no
Then, step (5c) is executed;
(5c) judges whether to have chosen all scale feature points in scale feature point sequence, if so, executing step (5d), otherwise
It executes step (5a);
All non-characteristic circles of crossing the border are formed non-characteristic circle sequence of crossing the border by (5d);
(6) non-overlap characteristic circle sequence is obtained:
(6a) chooses a non-characteristic circle of crossing the border from non-characteristic circle sequence of crossing the border, and selected non-characteristic circle of crossing the border is put into non-
In overlapping feature circle sequence;
(6b) chooses a non-characteristic circle of crossing the border from non-characteristic circle sequence of crossing the border, judge it is selected it is non-cross the border characteristic circle whether
It overlaps on disc with non-overlap characteristic circle sequence, if so, giving up the non-characteristic circle of crossing the border, otherwise, executes step
(6c);
Non- characteristic circle of crossing the border is put into non-overlap characteristic circle sequence by (6c);
(6d) judges whether to have chosen all non-characteristic circles of crossing the border in non-characteristic circle sequence of crossing the border, if so, step (7) are executed,
It is no to then follow the steps (6b);
(7) it is embedded in watermark:
(7a) chooses a non-overlap characteristic circle from non-overlap characteristic circle sequence, calculates the interior of selected non-overlap characteristic circle
Connect square;
(7b) according to the following formula, is embedded in watermark in the inscribed square of selection, obtains the inscribed square containing watermark:
G=embedWM (Q, W)
Wherein, G indicates that the inscribed square containing watermark, embedWM indicate that watermark embedding operation, Q indicate non-overlap characteristic circle
Inscribed square, W indicate the watermark of quasi- insertion;
(7c) judges whether to have chosen all non-overlap characteristic circles in non-overlap characteristic circle sequence, if so, obtaining aqueous impression
Otherwise picture executes step (7a);
The specific steps of the watermark extraction process are as follows:
(8) characteristic image sampling to be detected:
Image to be detected is extracted one piece of V ' × V ' pixel square matrix from upper left side by (8a), this square matrix is divided into
Size is 2 × 2 pixels and the sub-block that does not overlap, sorts left to right to obtain image to be detected block sequence;Wherein, V '=2 × d, d are
MeetUnder the conditions of maximum positive integer,Indicate downward floor operation, min expression is minimized
Operation, M ' and N ' respectively indicate the line number and columns of image to be detected;
(8b) cannot divided remainder after abandoning piecemeal;
(8c) successively chooses each sub-block in image to be detected block sequence, calculates sub-block in selected image to be detected block sequence
Average pixel value obtains mean pixel value sequence A ' to be detected;
(8d) is converted into characteristic image H ' to be detected according to the following formula, by mean pixel value sequence A ' to be detected;
H ' (μ, υ)=A 'β
Wherein, H ' (μ, υ) indicates the pixel value of μ row υ column in characteristic image H to be detected, A 'βIndicate mean pixel to be detected
The β value in value sequence A ', β=μ × d+ υ-d, d indicate satisfactionUnder the conditions of maximum it is just whole
Number,Indicate downward floor operation, min expression is minimized operation, and M ' and N ' respectively indicate the line number and column of image to be detected
Number, 1≤μ≤d, 1≤υ≤d;
(9) according to the following formula, initial characteristics point to be detected is extracted from characteristic image to be detected, obtains initial characteristics point sequence to be detected
Column:
F '=harrisLaplace (H ')
Wherein, F ' expression initial characteristics point sequence to be detected, harrisLaplace indicate to carry out Harris-Laplace operator meter
Calculate operation, H ' expression characteristic image to be detected;
(10) extension feature point sequence to be detected is calculated:
Each characteristic point in initial characteristics point sequence to be detected is successively chosen, it will be in selected initial characteristics point sequence to be detected
The corresponding line position of characteristic point set and expand as original twice with column position, obtain extension feature point sequence to be detected;
(11) scale feature point sequence to be detected is obtained:
(11a) successively chooses each characteristic point in extension feature point sequence to be detected, calculates selected extension feature to be detected
The corresponding scale of characteristic point in point sequence obtains characteristic point scaling sequence to be detected;
(11b) is in scale threshold interval [T1,T2] in, all scales met in characteristic point scaling sequence to be detected are chosen,
In, 1≤T1< T2≤16;
(11c) is successively using the corresponding extension feature point to be detected of the scale in characteristic point scaling sequence to be detected as scale feature
Point obtains scale feature point sequence to be detected;
(12) non-characteristic circle sequence of crossing the border to be detected is obtained:
(12a) chooses a scale feature point from scale feature point sequence to be detected, with the position where selected characteristic point
It is set to the center of circle, justifies by radius work of the product of the scale of characteristic point and given radius scale factor, crosses the border using the circle as non-
Characteristic circle;
(12b) judges whether non-characteristic circle of crossing the border is more than the boundary of image to be detected, if so, give up the non-characteristic circle of crossing the border,
Otherwise, step (12c) is executed;
(12c) judges whether to have chosen all scale feature points in scale feature point sequence to be detected, if so, executing step
(12d), it is no to then follow the steps (12a);
All non-characteristic circles of crossing the border are formed non-characteristic circle sequence of crossing the border to be detected by (12d);
(13) non-overlap characteristic circle sequence to be detected is obtained:
(13a) chooses a non-characteristic circle of crossing the border from non-characteristic circle sequence of crossing the border to be detected, by selected non-feature of crossing the border
Circle is put into non-overlap characteristic circle sequence to be detected;
(13b) chooses a non-characteristic circle of crossing the border from non-characteristic circle sequence of crossing the border to be detected, judges the selected non-spy that crosses the border
Whether sign circle overlaps on disc with non-overlap characteristic circle sequence to be detected, if so, give up the non-characteristic circle of crossing the border, it is no
Then, step (13c) is executed;
Non- characteristic circle of crossing the border is put into non-overlap characteristic circle sequence to be detected by (13c);
(13d) judges whether to have chosen all non-characteristic circles of crossing the border in non-characteristic circle sequence of crossing the border to be detected, if so, executing step
Suddenly (14), it is no to then follow the steps (13b);
(14) watermark is extracted:
(14a) chooses a non-overlap characteristic circle from non-overlap characteristic circle sequence to be detected, and it is special to calculate selected non-overlap
Levy the inscribed square of circle;
(14b) according to the following formula, extracts watermark in the inscribed square of selection, obtains the inscribed square containing watermark:
(G ', W ')=extractWM (Q ')
Wherein, the inscribed square after G ' expression reduction, the watermark after W ' expression reduction, extractWM indicate watermark extracting behaviour
Make, the inscribed square of Q ' expression non-overlap characteristic circle;
(14c) judges whether to have chosen all non-overlap characteristic circles in non-overlap characteristic circle sequence to be detected, if so, being gone back
Otherwise image after original executes step (14a).
2. the fragile reversible water mark method according to claim 1 based on image local area, it is characterised in that: step
The average pixel value of sub-block is realized according to following formula in the selected carrier image block sequence of calculating described in (1c):
Ai=(Bi1+Bi2+Bi3+Bi4)/4
Wherein, AiIndicate the average pixel value of i-th of sub-block in carrier image block sequence, Bi1It indicates the in carrier image block sequence
The pixel value being located at upper left position in i sub-block, Bi2It indicates to be located at the lower left corner in carrier image block sequence in i-th of sub-block
Pixel value at position, Bi3Indicate the pixel value for being located at upper right angular position in carrier image block sequence in i-th of sub-block, Bi4
It indicates to be located at the pixel value at lower right position in carrier image block sequence in i-th of sub-block.
3. the fragile reversible water mark method according to claim 1 based on image local area, it is characterised in that: step
The corresponding scale of characteristic point in selected extension feature point sequence is calculated described in (4a) to be realized according to following formula:
Sj=getScale (Uj)
Wherein, SjIndicate that j-th of scale in characteristic point scaling sequence, getScale indicate dimension calculation operation, UjIndicate that extension is special
Levy j-th of extension feature point in point sequence.
4. the fragile reversible water mark method according to claim 1 based on image local area, it is characterised in that: step
The inscribed square that selected non-overlap characteristic circle is calculated described in (7a) is realized according to following formula:
Wherein, L1Indicate the inscribed square upper left angle point of non-overlap characteristic circle in carrier image institute horizontal seat corresponding in the plane
Scale value, L2Indicate the inscribed square upper left angle point of non-overlap characteristic circle in carrier image institute ordinate corresponding in the plane
Value, D1Indicate the inscribed square bottom right angle point of non-overlap characteristic circle in carrier image institute abscissa value corresponding in the plane, D2
The inscribed square bottom right angle point of non-overlap characteristic circle is indicated in carrier image institute ordinate value corresponding in the plane, x is indicated
The center of circle of non-overlap characteristic circle indicates the center of circle of non-overlap characteristic circle in carrier image institute abscissa value corresponding in the plane, y
In carrier image institute ordinate value corresponding in the plane, E indicates the side of the inscribed square of non-overlap characteristic circle,Wherein,Indicate that downward floor operation, R indicate the radius of selected non-overlap characteristic circle.
5. the fragile reversible water mark method according to claim 1 based on image local area, it is characterised in that: step
The average pixel value of sub-block is realized according to following formula in the selected image to be detected block sequence of calculating described in (8c):
A′α=(B 'α1+B′α2+B′α3+B′α4)/4
Wherein, A 'αIndicate the average pixel value of the α sub-block in image to be detected block sequence, B 'α1Indicate image to be detected block sequence
The pixel value being located in the α sub-block at upper left position in column, B 'α2It indicates in image to be detected block sequence in the α sub-block
Positioned at the pixel value of lower-left angular position, B 'α3It indicates to be located at upper right angular position in image to be detected block sequence in the α sub-block
Pixel value, B 'α4It indicates to be located at the pixel value at lower right position in image to be detected block sequence in the α sub-block.
6. the fragile reversible water mark method according to claim 1 based on image local area, it is characterised in that: step
It is real according to following formula that the corresponding scale of characteristic point in selected extension feature point sequence to be detected is calculated described in (11a)
Existing:
S′ε=getScale (U 'ε)
Wherein, S 'εIndicate that the ε feature point scale in characteristic point scaling sequence to be detected, getScale indicate dimension calculation behaviour
Make, U 'εIndicate the ε extension feature point in extension feature point sequence to be detected.
7. the fragile reversible water mark method according to claim 1 based on image local area, it is characterised in that: step
The inscribed square that selected non-overlap characteristic circle is calculated described in (14a) is realized according to following formula:
Wherein, L '1Indicate the inscribed square upper left angle point of non-overlap characteristic circle in image to be detected institute cross corresponding in the plane
Coordinate value, L '2Indicate that the inscribed square upper left angle point of non-overlap characteristic circle is corresponding in the plane vertical in image to be detected institute
Coordinate value, D '1Indicate the inscribed square bottom right angle point of non-overlap characteristic circle in image to be detected institute cross corresponding in the plane
Coordinate value, D '2Indicate that the inscribed square bottom right angle point of non-overlap characteristic circle is corresponding in the plane vertical in image to be detected institute
Coordinate value, in image to be detected institute abscissa value corresponding in the plane, y ' expression is non-in the center of circle of x ' expression non-overlap characteristic circle
The center of circle of overlapping feature circle in carrier image to be detected institute ordinate value corresponding in the plane, E ' expression non-overlap characteristic circle
The side of square is inscribed,Wherein,Indicate downward floor operation, non-overlap selected by R ' expression is special
Levy the radius of circle.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610333977.4A CN106023055B (en) | 2016-05-19 | 2016-05-19 | Fragile reversible water mark method based on image local area |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610333977.4A CN106023055B (en) | 2016-05-19 | 2016-05-19 | Fragile reversible water mark method based on image local area |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106023055A CN106023055A (en) | 2016-10-12 |
CN106023055B true CN106023055B (en) | 2019-03-08 |
Family
ID=57094938
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610333977.4A Active CN106023055B (en) | 2016-05-19 | 2016-05-19 | Fragile reversible water mark method based on image local area |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106023055B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108053359A (en) * | 2017-12-07 | 2018-05-18 | 北京泛融科技有限公司 | A kind of block chain embedding grammar of wavelet field watermark in image |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102156955A (en) * | 2011-04-11 | 2011-08-17 | 西安电子科技大学 | Robust reversible watermark embedding and extracting method based on histogram neighborhood |
CN104156909A (en) * | 2014-09-05 | 2014-11-19 | 西安电子科技大学 | Robust reversible watermark embedding and detecting method on basis of image content |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20060239500A1 (en) * | 2005-04-20 | 2006-10-26 | Meyer Thomas W | Method of and apparatus for reversibly adding watermarking data to compressed digital media files |
-
2016
- 2016-05-19 CN CN201610333977.4A patent/CN106023055B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102156955A (en) * | 2011-04-11 | 2011-08-17 | 西安电子科技大学 | Robust reversible watermark embedding and extracting method based on histogram neighborhood |
CN104156909A (en) * | 2014-09-05 | 2014-11-19 | 西安电子科技大学 | Robust reversible watermark embedding and detecting method on basis of image content |
Non-Patent Citations (2)
Title |
---|
ROI–based Tamper Detection and Recovery for Medical Images Using Reversible Watermarking Technique;Osamah Al-qershi等;《International Conference on Information Theory and Information Security (ICITIS)》;20101231;全文 |
ROI和轮廓波结合的医学图像盲水印算法;李文娜等;《哈尔滨工程大学学报》;20130731;第34卷(第7期);全文 |
Also Published As
Publication number | Publication date |
---|---|
CN106023055A (en) | 2016-10-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhu et al. | Bag-of-visual-words scene classifier with local and global features for high spatial resolution remote sensing imagery | |
CN101520894B (en) | Method for extracting significant object based on region significance | |
Wang et al. | A robust blind color image watermarking in quaternion Fourier transform domain | |
CN103116763B (en) | A kind of living body faces detection method based on hsv color Spatial Statistical Character | |
CN109492416A (en) | A kind of guard method of big data image and system based on safety zone | |
CN101593272B (en) | Human face feature positioning method based on ASM algorithm | |
CN105825183B (en) | Facial expression recognizing method based on partial occlusion image | |
CN109753914A (en) | A kind of license plate character recognition method based on deep learning | |
CN101930549B (en) | Second generation curvelet transform-based static human detection method | |
CN107886062B (en) | Image processing method, system and server | |
CN102194208B (en) | Image falsification detecting and falsification positioning method based on image signature | |
CN103984966A (en) | SAR image target recognition method based on sparse representation | |
CN104915676A (en) | Deep-level feature learning and watershed-based synthetic aperture radar (SAR) image classification method | |
CN106446772A (en) | Cheating-prevention method in face recognition system | |
CN109948593A (en) | Based on the MCNN people counting method for combining global density feature | |
TW200929008A (en) | Multi-direction human face detection method | |
CN100455267C (en) | Method for recording identification of fetus sexes through B ultrasonic | |
CN104680545B (en) | There is the detection method of well-marked target in optical imagery | |
CN106056523B (en) | Blind checking method is distorted in digital picture splicing | |
CN102332084A (en) | Identity identification method based on palm print and human face feature extraction | |
CN104519361A (en) | Video steganography analysis method based on space-time domain local binary pattern | |
CN105138951B (en) | Human face portrait-photo array the method represented based on graph model | |
CN111144314A (en) | Method for detecting tampered face video | |
CN105894037A (en) | Whole supervision and classification method of remote sensing images extracted based on SIFT training samples | |
CN109829519A (en) | Classifying Method in Remote Sensing Image and system based on adaptive space information |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right |
Effective date of registration: 20210721 Address after: 400000 unit 1, building 1, phase 3, R & D building, Xiyong micro power park, Shapingba District, Chongqing Patentee after: Chongqing Institute of integrated circuit innovation Xi'an University of Electronic Science and technology Address before: 710071 No. 2 Taibai South Road, Shaanxi, Xi'an Patentee before: XIDIAN University |
|
TR01 | Transfer of patent right |