CN108805788A - Reversible water mark method based on image topology structure - Google Patents
Reversible water mark method based on image topology structure Download PDFInfo
- Publication number
- CN108805788A CN108805788A CN201810492373.3A CN201810492373A CN108805788A CN 108805788 A CN108805788 A CN 108805788A CN 201810492373 A CN201810492373 A CN 201810492373A CN 108805788 A CN108805788 A CN 108805788A
- Authority
- CN
- China
- Prior art keywords
- image
- block
- histogram
- watermark
- information
- 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
-
- 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
- G06T2201/0203—Image watermarking whereby the image with embedded watermark is reverted to the original condition before embedding, e.g. lossless, distortion-free or invertible watermarking
Abstract
A kind of reversible water mark method based on image topology structure, belongs to field of multi-media information safety.The present invention relates to a kind of bianry image reversible water mark methods based on image topology structure.The reversible water mark method, as histogram abscissa, ensures topological structure invariance by using image topology structure alternate image pixel value, generates statistic histogram and is embedded in watermark information.The present invention is suitable for bianry image, improves the visual quality of bianry image containing watermark, can also be combined embedded double-level watermark information with existing gray level image reversible water mark algorithm, improve existing algorithm hidden capacity, increase the versatility of inventive algorithm.
Description
Technical field
The invention belongs to reversible water mark, information security, copyright authentication technical fields.
Background technology
With the wide-scale distribution of digital content, copyright and authentication question become more and more important.For this purpose, image watermarking and can
Inverse digital watermark comes into being, especially in terms of requiring Hi-Fi multimedia content, such as military satellite image, medicine
Image, legal argument image.Earliest reversible water mark technology is published in 1997 ([1]), classical spatial domain reversible water mark skill
Art includes lossless compression ([2]), difference expansion ([3], [4], [5], [6]), histogram translation ([7], [8]), predicts that error is poor
Value extension ([9], [10]) etc..Currently a popular reversible water mark research work is concentrated mainly on gray level image or coloured image,
These images have larger redundancy (pixel, which slightly changes, to be not easy to be noticeable), therefore Information hiding is relatively easy to.However it is right
In being widely used in for scanning or the binary pictures such as fax paper, black and white woodcut, handwritten signature, image redundancy it is less and
The slight change of partial pixel point can cause image fault, this is very challenging for reversible water mark technology.
It is existing visual without causing by the embedded hiding information of certain rule based on the reversible water mark of bianry image
Pay attention to.But these methods are often relatively complicated or image fault after Information hiding is larger.Ho etc. first passes through image boundary
It constructs a difference diagram and piecemeal (size 1x4) is carried out to difference diagram, watermark is then realized by some pixel in flipping block
The insertion [11] of information.Yi etc. has used lossless compression method to be compressed to obtain space to completely black complete white and parts of images block
Embedding data [12].
The existing reversible water mark method based on bianry image be often difficult to vision concealment effect, Information Hiding Capacity it
Between reach an ideal balance.In order to improve vision concealment effect, the pixel position of change is often reduced, then in this way
Information Hiding Capacity can be reduced.In order to improve Information Hiding Capacity, often have to compare original image big change, in
It is vision concealment effect to decline.How to balance the two well is a prodigious challenge.
Invention content
The present invention needs to solve following key technical problem:
It needs to propose a kind of method of new efficient reversible water mark, determines which block is suitble to mutually turn in image block
Change, enables to visual distortion as small as possible to there is ideal concealment effect;
Need to propose a kind of reversible information hide, detection, recovery scheme, there is hiding quickly, extraction hiding information
Speed;Distortionless image after the completion of information extraction;With ideal Information Hiding Capacity and vision concealment effect.
It needs to propose a kind of flexibly expansible Information hiding, detection scheme, can support bianry image, gray level image etc.
A variety of different-format images.
It needs to propose new Information hiding, detection scheme, known reversible water mark scheme can be compatible with, remain able to carry
High hiding information capacity, while the eye fidelity having had.
Technical scheme of the present invention
Reversible water mark method based on image topology structure is:Including the bianry image based on image topology structure can be against the current
Impression method and the bianry image reversible water mark method based on image topology structure and the calculation of existing space domain gray level image reversible water mark
The gray level image reversible water mark method based on image topology structure that method is combined;
Bianry image reversible water mark method based on image topology structure includes watermark telescopiny and watermark extracting and original
Beginning image recovery process
1st step, watermark embedding method, step are:
1) image block:Piecemeal is carried out to original image first, each piecemeal can be square, can also be square
Shape;
2) connection block collection generates and histogram generates:It chooses the hiding block that those are capable of embedding information and is incorporated to connection block collection;
The hiding block for capableing of embedding information refers to that monochrome pixels is more balanced in this piecemeal, black to carry out wherein
The overturning of white pixel will not cause too big vision attention;It is also required that the piecemeal for capableing of embedding information has afterwards before hiding
Topological structure invariance enables to visual distortion minimum in this way;Here topological structure invariance refer in block
After monochrome pixels overturning, the connectivity for the connected domain being derived by black pixel point is constant;Connection block collection is divided into two
A subset, calculate separately in corresponding subset can embedded block value, valueOne of subset is used for structure
The histogram HD for being worth the number of appearance using the value of the hiding block of embedding information as abscissa for ordinate is made, another subset is used
To construct histogram of the number for being worth appearance to be embedded in peak value zero to the value of the hiding block of information as abscissa for ordinate
HA;
3) peak value zero is embedded in information:Histogram HA is embedded in the peak value zero pair of histogram HA using parity check system
Information and zero dot position information;
4) watermark information is embedded in:Histogram HD utilizes histogram translation mode insertion watermark information and histogram HA abscissas
Concentrate the parity of black pixel point number in image block;
2nd step, watermark extracting and original image restoration methods, step are:
1) histogram generates:With step 1) in watermark telescopiny described in the 1st step, 2) identical mode is generated and is believed with embedded
Breath hiding block value be abscissa, value appearance number be the histogram HD of ordinate and to be embedded in peak value zero to information
Hide block value be abscissa, value appearance number be ordinate histogram HA;
2) peak value zero is to information extraction:Extraction histogram HA abscissas concentrate black pixel point in image block in order
Binary stream is switched to the decimal system and determines the peak value zero pair selected in histogram HD by several parities;
3) watermark information extracts:According to the 2nd step 2) in the histogram peak zero that determines to using the inverse translation side of histogram
Formula extracts watermark information, histogram HA abscissas concentrate the parity of black pixel point number and parts of images in image block extensive
It is multiple;
4) image restores:Utilize the 2nd step 3) in the histogram HA abscissas that extract concentrate black pixel point in image block
Several parities restores black pixel point in original picture block.
2) the connection block collection generates in above-mentioned 1st step and histogram generation specifically includes following sub-step:
2.1) histogram HD corresponds to connection subset original block and determines:According to the complete white position in the block of the random selected digital image of key
As starting point and image black picture element number in the block, black picture element number size approximate image block pixel total number is selected
1/2, the legal moving direction of starting point is randomly choosed according to key, is moved to legal position and is set to this position pixel black
Colour vegetarian refreshments, cycle is until black pixel point total number is initial selected digital image block black picture element number in image block;
2.2) histogram HD corresponds to connection subset and generates:It is incorporated to using 2.1) the middle initial connection block generated as current block
It is connected to block collection, connection block size is b1*b2, upsets sequence { 0,1,2,3 ..., b1*b2-1 } at random and generates new sequence S, according to sequence
Sequence in row S negates the pixel value of corresponding position in current block, and whether the block after judgement changes meets connection block and black
White pixel is more balanced in this piecemeal and does not concentrate two conditions of appearance in connection block, makees the block after change if meeting
It is incorporated to connection block collection for current block, pixel value is otherwise negated again and reverts to current block;
2.3) histogram HA corresponds to connection subset original block and determines:First is selected in lexcographical order is not belonging to HD correspondence connections
The image block of subset;Described image block meets connection, monochrome pixels distributing equilibrium, not in the corresponding connection subsets of histogram HD
Occur.
2.4) histogram HA corresponds to connection subset and generates:It is incorporated to using 2.3) the middle initial connection block generated as current block
It is connected to block collection, connection block size is b1*b2, upsets sequence { 0,1,2,3 ..., b1*b2-1 } at random and generates new sequence S, according to sequence
Sequence in row S negates the pixel value of corresponding position in current block, and whether the block after judgement changes meets connection block, black and white
Pixel in this piecemeal it is more balanced be not connected in subset three conditions of appearance histogram HD and HA are corresponding, if meeting
It is incorporated to connection block collection using the block after change as current block, pixel value is otherwise negated again and reverts to current block.
The above-mentioned reversible water mark method based on image topology structure can be with existing spatial domain gray level image reversible water mark
Algorithm is mutually compatible with, and improves existing reversible water mark algorithm hidden capacity, including watermark telescopiny and watermark extracting and image restore
Process.
1st step, watermark telescopiny, step are:
1) using existing, the gray level image reversible water mark algorithm insertion first layer watermark information generation based on spatial domain is aqueous
Print gray level image;
2) watermark gray level image will be contained and is mapped as bianry image, it can be against the current using the bianry image based on image topology structure
Impression method is embedded in second layer watermark information and generates bianry image containing watermark;
3) it is gray level image that will contain watermark bianry image inverse mapping.
2nd step, watermark extracting and image recovery method, step are:
1) gray level image is mapped as bianry image, uses the bianry image reversible water mark method based on image topology structure
Extraction watermark information simultaneously restores original bianry image;
2) it is gray level image by original bianry image inverse mapping, it can be against the current using the identical gray level image based on spatial domain
Print algorithm extraction watermark information simultaneously restores original image.
The advantages of the present invention:
1) a kind of scheme of efficient determining overturning monochrome pixels position is proposed, image before and after Information hiding can be kept
Piecemeal exports the topological invariance of figure, and payload does not occur to drop by a relatively large margin while improving visual quality containing watermarking images
It is low, to propose a kind of efficient reversible hidden algorithm;
2) Information hiding of proposition, extraction, recovery system have ideal vision concealment effect and ideal Information hiding
Capacity can reach a good balance between the two;
3) present invention can be to various forms of bianry images such as calligraphy work, woodcut, scan image, handwritten signature etc.
It carries out Information hiding extraction and image restores;A degree of copyright protection, copyright control, authentication role can be played;
4) present invention can be applied not only to bianry image, can also be applied to the common formats images such as gray level image;
5) present invention can be compatible with the existing gray level image reversible water mark method based on spatial domain, improve hidden capacity.
Description of the drawings
Fig. 1 is image block location of pixels figure of the present invention and 3 × 3 piecemeal exemplary plots;
Fig. 2 is the flow chart that the present invention carries out bianry image watermark insertion processing method M200;
Fig. 3 is the implementation exemplary plot of the present invention, includes two 3 × 3 subset starting blocks;
Fig. 4 is the algorithm 1 that the present invention generates histogram HD abscissa collection;
Fig. 5 is an implementation exemplary plot of the algorithm 1 that the present invention generates histogram HD abscissa collection
Fig. 6 is the flow chart of histogram translation algorithm 2;
Fig. 7 is the flow chart that the present invention carries out bianry image watermark extracting and image recovery method M700;
Fig. 8 is that the present invention one generates two subset few examples figures that connected set divides;
Fig. 9 be the present invention a histogram insertion state before and after exemplary plot;
Figure 10 is the insertion design sketch of the present invention;
Figure 11 is 3 flow chart of algorithm that the present invention program is applied to gray level image;
Figure 12 is the corresponding bianry image exemplary plot of the different binaryzation functions of four kinds of the present invention;
Figure 13 is the exemplary plot that the present invention combines gray level image histogram translation reversible water mark algorithm.
Figure 14 is the exemplary plot that the present invention combines gray level image prediction error difference extension reversible water mark algorithm.
Specific implementation mode
The specific implementation mode of the present invention is introduced below in conjunction with drawings and examples.
Embodiment 1
Bianry image reversible water mark method based on image topology structure, this method include:
First part:Bianry image reversible water mark insertion based on image topology structure
For ease of algorithm description, Fig. 1 give in the present invention through frequently with term and mark, and for 3 × 3 piecemeals
Example.In a piecemeal, if two black pixel points vertically adjacent to or it is horizontally adjacent, two black pixel points are known as
It is adjacent.Each black pixel point corresponds to a vertex, and two adjacent black pixel points set a line.Therefore one
Piecemeal can correspond to a figure.Boundary refers to the white pixel point set adjacent with black picture element point set.In addition, one
The weight of piecemeal is that black picture element therein is counted out.For the piecemeal of a n1 × n2, we divide successively since the upper left corner
With a serial number, 0,1 ..., n1 × n2-1, then the value of a piecemeal beAs shown in fig. 1, the piecemeal weight
Amount is 5, and boundary is { 0,2,3,5 }, and the value of the piecemeal is that 151. clearly the piecemeal for 3 × 3, piecemeal take [0,511] only
One value.The flow chart of the watermark telescopiny M200 of bianry image is as shown in Fig. 2.
1) constituent instruments (BMP) that bitmap format is obtained from binarized-image file, at this time there are one each pixels
Bit indicates its monochrome informatio.Assuming that its size of bianry image I is M*N pixels, the division of spatial domain is carried out to the image, point
At etc. the nonoverlapping fritter of scales its size be n1 × n2 (n1=n2=3).Notice that fritter needs not be square, Ke Yishi here
Rectangle, the present invention illustrate specific implementation mode by embodiment of 3*3.Borderline redundance is not handled.
2) it from several fritters after step 1) division, chooses a part of be used as and meets the requirements, it can be with the small of hiding information
Block.Here it can be referred to after carrying out the redirecting of monochrome pixels, will not visually be caused too big with the fritter of hiding information
Vision attention.We here it is considered that those much the same fritters of monochrome pixels number have ideal concealment property because from regarding
It is not easy to cause visual attention after overturning a monochrome pixels in feel.In addition redirecting for monochrome pixels should not destroy this fritter
The connectivity of middle black pixel point;Monochrome pixels redirect the Basic Topological that will not destroy black pixel point in this fritter.
Definition can be had the property that with the fritter of hiding information:
I. for the fritter of a n1 × n2, weight is about n1 × n2/2;
Ii. the figure derived from black pixel point should be connection in block;
The step for practical operation in, we only need to judge divided fritter value whether we construct can be hidden
In the histogram abscissa set for hiding information.
Selecting can be with the fritter of hiding information, and image block collection is divided into two disjoint subsets, and in each subset
Only only there are one the differences of pixel for two adjacent blocks.Subset division mode is as follows:
2.1) according to the connection block for generating one 3x3 of key generation, in this, as the starting block of one of subset.Fig. 3
It is two 3 × 3 subset starting block embodiments of the present invention.
2.2) Fig. 4 is 1 flow chart of algorithm that present invention generation can be connected to subset with hiding information piecemeal, and for 3 × 3
Piecemeal, according to algorithm 1 generate histogram abscissa example.In view of extracting the needs of information, it is desirable that after hiding information
Piecemeal must still can in the set of hiding information piecemeal, accordingly can from one can hiding information piecemeal export one and hide letter
The change chain of breath.One generation chain, step can be exported using method M400 with the set of hiding information piecemeal to generate
S410 chooses first can be using the piecemeal of hiding information as starting block;Then its weight is calculated in step S420;Next in step
Rapid S430 upsets pixel value change sequence { 0,1,2,3,4,5,6,7,8 } according to key and obtains new sequence, step S440 at random
According to newly-generated sequence overturn corresponding position pixel, it is black picture element that even current location, which is the overturning of white pixel point,
Point, weight add 1;If current pixel point is black pixel point, overturning is white pixel point, and weight subtracts 1;Followed by the new block of judgement
Whether meeting condition makes the topological structure of definition not change, if the block after changing is connected to and in set before
It never occurred, is then incorporated into set, otherwise pixel value negates the former block of recovery.The next position is jumped to, until finding
Terminate.Fig. 5 gives the exemplary plot generated using method M400.
Judge whether generating subset length is more than threshold value, terminate more than threshold value, being less than threshold value, then another key value increases 1 again certainly
It gos to step 2.1).
2.3) starting block of first connection block for being not belonging to generating subset as another subset is chosen from lexcographical order;
2.4) pixel change sequence { 0,1,2,3,4,5,6,7,8 } is upset at random to obtain new sequence according to key, presses
Random sequence overturns current pixel location color, and it is black pixel point that even current location, which is the overturning of white pixel point, and weight adds
1;If current pixel point is black pixel point, overturning is white pixel point, and weight subtracts 1.If the block after changing is connected to and at it
It was never occurred in preceding set, is then incorporated into set, and otherwise restored former block.The next position is jumped to, until finding
Terminate.
Judge whether generating subset length is more than threshold value, terminates more than threshold value, then sought again in lexcographical order less than threshold value
Next satisfactory connection block is looked for go to step again 2.4).Notice that the intersection of two subsets is sky.
3) the image block subset (having less block type) determined according to step 2) carries out Information hiding, record subset block
Parity, by the peak value zero of original histogram to being embedded in.Embedded mode is as follows:
I. whether the parity of decision block and wanted embedding information are identical, and the identical constant difference of maintainance block jumps to ii
Ii. a pixel makes the pattern of embedded block change into the pattern of the right adjacent block in flipping block
4) mode for using histogram to translate is embedded in watermark information and the affiliated block region weight of histogram HA abscissa collection is strange
Idol, Embedded step are as follows:
4.1) it uses the value of the image block subset of algorithm M400 generations as histogram abscissa, is gone out with the value of each pattern
Existing number generates corresponding histogram as ordinate;
4.2) several peak value zeros pair are selected in histogram, i.e., the peak more than some threshold value is found between two zeros
It is worth point, matches peak point and zero point in pairs if there are two peak points, between each other without intersection, if one and only one
Peak point then matches peak point and left side zero point, the right zero point as next minor peaks zero to the starting point of search, if
It is not present, then directly using the right zero point as next minor peaks zero to search starting point;
4.3) by the block in peak value zero point all in one unit of zero point direction translation, i.e., each block
One pixel of overturning makes its pattern become adjacent block pattern;
4.4) the embedded message in peak point remains unchanged if message is 0, be 1 by embedded block to close to zero side
To one unit of translation.
Second part:The extraction of bianry image reversible water mark and original image based on image topology structure restore
1) Fig. 7 is the flow chart that the present invention carries out bianry image watermark extracting and restoration methods.Start to read and contains watermark
Bianry image constructs two connection subsets and generates correspondence according to the value of two of generation connection subsets then to image block
Histogram is as follows:
I. use key according to method M400 construct one can hiding information block subset, if sub-set size is less than threshold value T
(present invention in T=70) enable key add 1 reuse method M400 constructions can hiding information block subset until sub-set size is big
In given threshold T, the present invention this subset is indicated using HDX;
Ii. first is found in lexcographical order is not belonging to HDX and the starting block of connection.Application method M400 generates one can
The subset for hiding additional information block, is indicated using HAX.It needs to meet condition in generating process:The new block connection generated, is not belonging to
HDX did not occurred in HAX.Therefore two sons generated integrate the condition for meeting intersection as empty set.
Iii. scan image piecemeal, using be connected to contained by subset HDX and HAX can embedded block value as histogram cross
The value occurrence number of coordinate, image block generates image statistics histogram as histogram ordinate.
Fig. 8 is a kind of few examples figure of HDX and HAX.
2) all image blocks for belonging to HAX are found, are arranged according to specified sequence, additional information is extracted, it is described additional
Information is the parity (for example odd number means to hide 1, even number 0) that black picture element is counted out in image block.By binary stream
Switch to peak value zero pair and zero dot position information that the decimal system obtains histogram HD;
3) according to the peak value zero extracted in step 2) to carrying out data extraction and recovery, and following two rule is followed:
When i. extracting data, if encountering the block extraction watermark information " 1 " positioned at peak point;If encountering peak point adds 1 (zero
Point is more than peak point)/extraction the watermark information " 0 " at (zero point be less than peak point) that subtracts 1;
Ii. after the completion of data are all extracted by the block in peak value and zero point section all to separate zero point side
To 1 unit of translation;Finally terminate in step S770.
In order to enable histogram translation algorithm is embedded in water it can be readily appreciated that Fig. 9 gives a histogram with 50% probability
The front and back schematic diagram changed of official seal breath, the peak value zero chosen in histogram is to for (P1, P3), (P5, P4), (P6, P8).
Figure 10 is the embedded design sketch of the present invention.Using 4 width images as example, 3 × 3 points are hidden based on what algorithm 2 generated
Set of blocks, for simplicity we tested on the data set containing 1000 width images using identical watermark information,
And 4 pictures are arbitrarily had chosen, show that the invention has ideal vision concealment effect, Information Hiding Capacity, first is classified as original
Beginning image, second is classified as aqueous watermark image, and third is classified as original image and disparity map containing watermarking images.PSNR is to be often used in weighing
The measurement of the visual perception of reconstructed image quality is measured, experiment shows that PSNR has ideal numerical value.
Embodiment 2
Bianry image reversible water mark method based on image topology structure is calculated with existing space domain gray level image reversible water mark
The gray level image reversible water mark method based on image topology structure that method is combined includes
First part:Gray level image reversible water mark insertion based on image topology structure
For ease of algorithm description, it is contemplated that width gray level image G its size is M*N pixels, wherein M and N be all 3 it is whole
Several times are not that the gray level image of 3 integral multiple carries out BORDER PROCESSING.Gray level image reversible water mark based on image topology structure is embedding
It is as shown in Fig. 11 to enter method flow diagram.
1) use histogram translation algorithm, difference expansion algorithm etc. based on prediction error embedded the in gray level image G
One layer of watermark information obtains the W of gray level image containing watermark;
2) binaryzation is carried out to image using invertible function f (x), wherein invertible function f (x) refers to function y=f (x)
There are inverse function x=f-1(y), the pixel p being expressed as during gray level image is mapped as bianry image in gray level imageI, j
There is unique bianry image pixel bI, jIt is corresponding to it, and bianry image pixel b during inverse mappingI, jCorresponding unique gray scale
Pixel p in imageI, j.In addition, also needing to meet condition when design invertible function f (x):The bianry image that mapping generates should have
Redundancy rather than Random Graph.Next the bianry image insertion second layer watermark information generated using mapping obtains two-value containing watermark
Image WB, mode are similar with M200.
Figure 12 is the corresponding bianry image exemplary plot of four kinds of difference binaryzation functions of the present invention program design.This programme is adopted
With 512x512 standard drawings lena as an example, the four kinds of binaryzation functions used are respectively:(a)f1(x)=pI, j, 7(i, j are indicated
Pixel position);(b)f2(x)=pI, j, 6;(c)
f3(x)=pI, j, 6+pI, j, 5*pI, j, 4+pI, j, 3*pI, j, 2+pI, j, l*pI, j, 0;
(d)f4(x)=pI, j, 7+pI, j, 6*pI, j, 5.It can be seen that (a), (b), (d) three kinds of binaryzation Function Mappings in Figure 10
There is the bianry image of generation larger redundancy can be used as binary map progress watermark insertion, and (c) generation binary map is superfluous similar to Random Graph
Remaining be approximately 0 be not suitable for carry out watermark insertion.
3) the inverse function x=f of y=f (x) is utilized-1(y) it is the ash containing double-level watermark by the WB inverse mappings of bianry image containing watermark
Degree figure.
The example that Figure 13 is the present invention program to be combined with classical histogram translation algorithm, in experiment first with
Existing histogram translation algorithm is embedded in first layer watermark information, and size is about 4800bits, reuses the present invention program and carries out the
Two layers of watermark information insertion are to increase Information Hiding Capacity, size 128bits.Using standard drawing Lena as test chart, greatly
Small is 512x512.It is respectively according to sequence from left to right:Original image Lena;Carry out the once histogram based on gray level image
The insertion result figure of figure translation schemes;It is carried out using figure binaryzation function in the result figure translated based on gray level image histogram
The bianry image of image binaryzation;M800 scenario outcomes figures are used in the result figure translated based on gray level image histogram.It presses
It is respectively according to the binaryzation function that sequence from top to bottom uses:
(1)f1(x)=pI, j, 7(i, j indicate pixel position);
(2)f2(x)=pI, j, 6;
(3)f4(x)=pI, j, 7+pI, j, 6*pI, j, 5。
Figure 14 is the example that the present invention program is combined with classical prediction error difference expansion algorithm, first in experiment
It is embedded in first layer watermark information first with existing prediction error difference expansion algorithm, size is about 4800bits, reuses this hair
Bright scheme carries out the insertion of second layer watermark information to increase Information Hiding Capacity, size 128bits.Made using standard drawing Lena
For test chart, size 512x512.It is respectively according to sequence from left to right:Original image Lena;It carries out once being based on gray scale
The insertion result figure of the histogram translation schemes of image;Figure two-value is used in the result figure translated based on gray level image histogram
Change the bianry image that function carries out image binaryzation;M800 schemes are used in the result figure translated based on gray level image histogram
Result figure.The binaryzation function used in accordance with the order from top to bottom is respectively:(1)f1(x)=pI, j, 7(i, j indicate pixel
Position);(2)f2(x)=pI, j, 6;(3)f4(x)=pI, j, 7+pI, j, 6*pI, j, 5。
Second part:The extraction of gray level image reversible water mark and original image based on image topology structure restore
1) gray level image containing double-level watermark is mapped as bianry image containing watermark using function y=f (x), using based on
The bianry image reversible water mark extraction of image topology structure and image recovery method M700 extraction watermark informations simultaneously restore original two
It is worth image;
2) the inverse function x=f of y=f (x) is utilized-1(y) it is gray level image containing watermark by original bianry image inverse mapping, makes
With identical gray level image reversible water mark algorithm (histogram translation algorithm) the extraction watermark information based on spatial domain and restore former
Beginning image.
Bibliography
[1]J.M.Barton,``Method and Apparatus for Embedding Authentication
Information Within Digital Data",U.S.Patent.5646997,1997.
[2]M.U.CelikMehmet,G.Sharma,andA.M.Tekalp,``Reversible data hiding",
Proc.IEEE ICIP,2002.
[3]J.Tian,``Reversible data embedding using a difference expansion",
IEEE Trans.Circuits Syst.Video Technol.vol.13,no.8,pp.890-896,Aug.2003.
[4]I.Dragoi,and D.Coltuc,``Local-Prediction-Based Difference
Expansion Reversible Watermarking",IEEE Trans.on Image Processing,vol.23,
Pp.1779-1790, Apr.~2014.
[5]A.M.Alattar,``Reversible watermark using difference expansion of
triplets",in Proc.IEEE ICIP.vol.1,Barcelona Spain,2003,pp.501-504.
[6]A.M.Alattar,``Reversible watermark using difference expansion of
quads",in Proc.IEEE ICASSP.vol.3,pp.377-380,Toulouse,France,2004.
[7]I.Dragoi,``Adaptive Pairing Reversible Watermarking",inIEEE
Trans.on Image Processing.vol.25,pp.2420-2422,2016.
[8]Z.Ni,Y.Shi,N.Ansari,and W.Su,``Reversible data hiding”,IEEE
Trans.Circuits Syst.Video Technol,vol.16,pp.354-362,2006.
[9]Thodi D M,Rodriguez J J.“Prediction-error based reversible
watermarking”.International Conference on Image Processing,2004:1549-
1552Vol.3.
[10]Luo L,Chen Z,Chen M,et al.“Reversible image watermarking using
interpolationtechnique”.IEEE Transactions on Information Forensics and
Security,2010,5(1):187-193.
[11]Ho,Y.A.,Chan,Y.K.,Wu,H.C.and Chu,Y.P,``High-capacity reversible
data hiding in binary images using pattern substitution",Computer Standards
and Interfaces.vol.31,pp.787-794,2009.
[12]ShuangYi,YicongZhou,``Binary-block embedding for reversible data
hiding in encrypted images”,SignalProcessing,volume 133,April 2017,Pages 40-
51.
Claims (3)
1. a kind of reversible water mark method based on image topology structure, it is characterised in that:Include two based on image topology structure
The watermark telescopiny and watermark extracting and original image recovery process of value image reversible water mark;
1st step, watermark telescopiny include:
1) image block:Piecemeal is carried out to original image first, each piecemeal can be square, can also be rectangle;
2) connection block collection generates and histogram generates:It chooses the hiding block that those are capable of embedding information and is incorporated to connection block collection;It is described
The hiding block for capableing of embedding information refers to that monochrome pixels is more balanced in this piecemeal, to carry out black and white picture wherein
The overturning of element will not cause too big vision attention;It is also required that the piecemeal for capableing of embedding information has topology afterwards before hiding
Structure invariance enables to visual distortion minimum in this way;Here the black and white in block that topological structure invariance refers to
After pixel overturning, the connectivity for the connected domain being derived by black pixel point is constant;Connection block collection is divided into two sons
Collection, calculate separately in corresponding subset can embedded block value, valueOne of subset be used for construct with
The value of the hiding block of embedding information is abscissa, and the number for being worth appearance is the histogram HD of ordinate, another subset is used for structure
Make the histogram HA for being worth the number of appearance to be embedded in peak value zero to the value of the hiding block of information as abscissa for ordinate;
3) peak value zero is embedded in information:Histogram HA is using the peak value zero of parity check system insertion histogram HA to information
And zero dot position information;
4) watermark information is embedded in:Histogram HD is concentrated using histogram translation mode insertion watermark information and histogram HA abscissas
The parity of black pixel point number in image block;
2nd step, watermark extracting and original image recovery process include:
1) histogram generates:With step 1) in watermark telescopiny described in the 1st step, 2) identical mode is generated with embedding information
Hide block value be abscissa, value appearance number be the histogram HD of ordinate and information hidden with being embedded in peak value zero
The value of block be abscissa, value occur number be ordinate histogram HA;
2) peak value zero is to information extraction:Extraction histogram HA abscissas concentrate black pixel point number in image block in order
Binary stream is switched to the decimal system and determines the peak value zero pair selected in histogram HD by parity;
3) watermark information extracts:According to the 2nd step 2) in determine histogram peak zero to using histogram to be carried against translation mode
Take watermark information, histogram HA abscissas that the parity of black pixel point number and parts of images in image block is concentrated to restore;
4) image restores:Utilize the 2nd step 3) in the histogram HA abscissas that extract concentrate black pixel point number in image block
Parity restores black pixel point in original picture block.
2. the reversible water mark method as described in claim 1 based on image topology structure, which is characterized in that 2) institute in the 1st step
It states the generation of connection block collection and histogram generation specifically includes following sub-step:
2.1) histogram HD corresponds to connection subset original block and determines:According to the complete white position conduct in the block of the random selected digital image of key
Starting point and image black picture element number in the block select black picture element number size approximate image block pixel total number 1/2,
The legal moving direction that starting point is randomly choosed according to key, is moved to legal position and this position pixel is set to black picture
Vegetarian refreshments, cycle is until black pixel point total number is initial selected digital image block black picture element number in image block;
2.2) histogram HD corresponds to connection subset and generates:Using 2.1) the middle initial connection block generated connection is incorporated to as current block
Block collection, connection block size are b1*b2, upset sequence { 0,1,2,3 ..., b1*b2-1 } at random and generate new sequence S, according to sequence S
In sequence the pixel value of corresponding position in current block is negated, whether the block after judgement changes meets connection block and black and white picture
Element is more balanced in this piecemeal and is not concentrated in connection block and two conditions occurs, if meeting using the block after change as working as
Preceding piece is incorporated to connection block collection, otherwise negates pixel value again and reverts to current block;
2.3) histogram HA corresponds to connection subset original block and determines:First is selected in lexcographical order is not belonging to HD correspondence connection subsets
Image block;Described image block meets connection, and monochrome pixels distributing equilibrium does not go out in the corresponding connection subsets of histogram HD
It is existing;
2.4) histogram HA corresponds to connection subset and generates:Using 2.3) the middle initial connection block generated connection is incorporated to as current block
Block collection, connection block size are b1*b2, upset sequence { 0,1,2,3 ..., b1*b2-1 } at random and generate new sequence S, according to sequence S
In sequence the pixel value of corresponding position in current block is negated, judgement change after block whether meet connection block, black and white picture
Element in this piecemeal it is more balanced be not connected in subset three conditions of appearance histogram HD and HA are corresponding, will if meeting
Block after change is incorporated to connection block collection as current block, otherwise negates pixel value again and reverts to current block.
3. the reversible water mark method as described in claim 1 based on image topology structure, it is characterised in that:It is described based on figure
As the bianry image reversible water mark method of topological structure can be mutually compatible with existing spatial domain gray level image reversible water mark algorithm, carry
High existing reversible water mark algorithm hidden capacity, including watermark telescopiny and watermark extracting and image recovery process;
1st step, watermark telescopiny include:
1) using existing, the gray level image reversible water mark algorithm insertion first layer watermark information based on spatial domain is generated containing watermark ash
Spend image;
2) watermark gray level image will be contained and is mapped as bianry image, and use the bianry image reversible water mark side based on image topology structure
Method is embedded in second layer watermark information and generates bianry image containing watermark;
3) it is gray level image that will contain watermark bianry image inverse mapping;
2nd step, watermark extracting and image recovery process include:
1) gray level image is mapped as bianry image, is extracted using the bianry image reversible water mark method based on image topology structure
Watermark information simultaneously restores original bianry image;
2) it is gray level image by original bianry image inverse mapping, is calculated using the identical gray level image reversible water mark based on spatial domain
Method extracts watermark information and restores original image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810492373.3A CN108805788B (en) | 2018-05-22 | 2018-05-22 | Reversible watermarking method based on image topological structure |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810492373.3A CN108805788B (en) | 2018-05-22 | 2018-05-22 | Reversible watermarking method based on image topological structure |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108805788A true CN108805788A (en) | 2018-11-13 |
CN108805788B CN108805788B (en) | 2022-04-26 |
Family
ID=64092711
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810492373.3A Active CN108805788B (en) | 2018-05-22 | 2018-05-22 | Reversible watermarking method based on image topological structure |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108805788B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111127288A (en) * | 2019-12-16 | 2020-05-08 | 中移(杭州)信息技术有限公司 | Reversible image watermarking method and device and computer readable storage medium |
CN113112392A (en) * | 2020-01-13 | 2021-07-13 | 国际关系学院 | Watermark embedding and extracting method, embedding and extracting device, and processing system |
CN113205445A (en) * | 2021-04-16 | 2021-08-03 | 北京大学 | Reversible plain watermark method and device based on data compression |
CN114037593A (en) * | 2021-10-29 | 2022-02-11 | 淮阴工学院 | Reversible image watermarking algorithm based on reverse histogram translation |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040046774A1 (en) * | 2001-03-05 | 2004-03-11 | Rhoads Geoffrey B. | Image management system and methods using digital watermarks |
WO2004102464A2 (en) * | 2003-05-08 | 2004-11-25 | Digimarc Corporation | Reversible watermarking and related applications |
CN101651837A (en) * | 2009-09-10 | 2010-02-17 | 北京航空航天大学 | Reversible video frequency watermark method based on interframe forecast error histogram modification |
CN104268824A (en) * | 2014-09-19 | 2015-01-07 | 柳州师范高等专科学校 | Self-adaptation reversible watermarking method used for medical image authentication and doctor-patient information hiding |
CN105427229A (en) * | 2015-11-11 | 2016-03-23 | 九江学院 | Method for embedding reversible watermarks in duplex domains |
CN105447809A (en) * | 2015-12-07 | 2016-03-30 | 中国矿业大学 | Weed model based encrypted domain image reversible watermarking method |
CN106097236A (en) * | 2016-04-29 | 2016-11-09 | 西安电子科技大学 | Frequency domain robust image reversible water mark method based on Non-negative Matrix Factorization |
CN106803228A (en) * | 2016-12-07 | 2017-06-06 | 华北电力大学(保定) | Based on obtuse angle Forecasting Methodology, system that reversible information is hidden |
-
2018
- 2018-05-22 CN CN201810492373.3A patent/CN108805788B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040046774A1 (en) * | 2001-03-05 | 2004-03-11 | Rhoads Geoffrey B. | Image management system and methods using digital watermarks |
WO2004102464A2 (en) * | 2003-05-08 | 2004-11-25 | Digimarc Corporation | Reversible watermarking and related applications |
CN101651837A (en) * | 2009-09-10 | 2010-02-17 | 北京航空航天大学 | Reversible video frequency watermark method based on interframe forecast error histogram modification |
CN104268824A (en) * | 2014-09-19 | 2015-01-07 | 柳州师范高等专科学校 | Self-adaptation reversible watermarking method used for medical image authentication and doctor-patient information hiding |
CN105427229A (en) * | 2015-11-11 | 2016-03-23 | 九江学院 | Method for embedding reversible watermarks in duplex domains |
CN105447809A (en) * | 2015-12-07 | 2016-03-30 | 中国矿业大学 | Weed model based encrypted domain image reversible watermarking method |
CN106097236A (en) * | 2016-04-29 | 2016-11-09 | 西安电子科技大学 | Frequency domain robust image reversible water mark method based on Non-negative Matrix Factorization |
CN106803228A (en) * | 2016-12-07 | 2017-06-06 | 华北电力大学(保定) | Based on obtuse angle Forecasting Methodology, system that reversible information is hidden |
Non-Patent Citations (2)
Title |
---|
李立宗: "基于公钥的可逆数字水印", 《计算机应用》 * |
项洪印: "基于平滑分块基像素定位的可逆信息隐藏方法", 《计算机工程》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111127288A (en) * | 2019-12-16 | 2020-05-08 | 中移(杭州)信息技术有限公司 | Reversible image watermarking method and device and computer readable storage medium |
CN111127288B (en) * | 2019-12-16 | 2023-07-04 | 中移(杭州)信息技术有限公司 | Reversible image watermarking method, reversible image watermarking device and computer readable storage medium |
CN113112392A (en) * | 2020-01-13 | 2021-07-13 | 国际关系学院 | Watermark embedding and extracting method, embedding and extracting device, and processing system |
CN113112392B (en) * | 2020-01-13 | 2023-05-19 | 国际关系学院 | Watermark embedding and extracting method, watermark embedding and extracting device and processing system |
CN113205445A (en) * | 2021-04-16 | 2021-08-03 | 北京大学 | Reversible plain watermark method and device based on data compression |
CN113205445B (en) * | 2021-04-16 | 2023-10-17 | 北京大学 | Reversible clear watermark method and device based on data compression |
CN114037593A (en) * | 2021-10-29 | 2022-02-11 | 淮阴工学院 | Reversible image watermarking algorithm based on reverse histogram translation |
Also Published As
Publication number | Publication date |
---|---|
CN108805788B (en) | 2022-04-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108805788A (en) | Reversible water mark method based on image topology structure | |
Yang et al. | Improving histogram-based reversible data hiding by interleaving predictions | |
Wu et al. | A steganographic method for images by pixel-value differencing | |
EP3308371B1 (en) | System and method for digital watermarking | |
Yang et al. | Reversible steganography based on side match and hit pattern for VQ-compressed images | |
JP2000163594A (en) | Image pattern detecting method and device | |
CN110232651B (en) | Reversible information hiding method based on block prediction and differential value histogram translation | |
CN107993208A (en) | It is a kind of based on sparse overlapping group prior-constrained non local full Variational Image Restoration method | |
JPH10143604A (en) | Device for extracting pattern | |
CN111898136A (en) | High-fidelity reversible information hiding method and device based on prediction error value sequencing | |
Xia et al. | Towards deepfake video forensics based on facial textural disparities in multi-color channels | |
Goli et al. | Introducing a new method robust against crop attack in digital image watermarking using two-step sudoku | |
Wu et al. | Reversible data hiding with brightness preserving contrast enhancement by two-dimensional histogram modification | |
Chuang et al. | Joint index coding and reversible data hiding methods for color image quantization | |
EP0516477B1 (en) | Compression enhancement in graphics systems | |
Cu et al. | A robust watermarking approach for security issue of binary documents using fully convolutional networks | |
JP3514050B2 (en) | Image processing device | |
CN110738592A (en) | High-capacity reversible image watermarking algorithm based on multi-scale decomposition and interpolation expansion | |
CN109859090A (en) | Reversible water mark method and device based on human visual system | |
CN106658021A (en) | Method for embedding and detecting two types of watermarks of MPEG video | |
Yang et al. | A Data Hiding Method Based on Partition Variable Block Size with Exclusive-or Operation on Binary Image | |
CN117974409A (en) | Model training method, image processing method and related equipment | |
Dhavale et al. | High capacity, robust lossless EPR data hiding using CDCS with ROI tamper detection | |
CN117635411B (en) | Digital watermark processing method based on mixed domain decomposition technology | |
Wang et al. | A novel framework for object removal from digital photograph |
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 |