CN102903076A - Method for embedding and extracting reversible watermark of digital image - Google Patents
Method for embedding and extracting reversible watermark of digital image Download PDFInfo
- Publication number
- CN102903076A CN102903076A CN2012104094507A CN201210409450A CN102903076A CN 102903076 A CN102903076 A CN 102903076A CN 2012104094507 A CN2012104094507 A CN 2012104094507A CN 201210409450 A CN201210409450 A CN 201210409450A CN 102903076 A CN102903076 A CN 102903076A
- Authority
- CN
- China
- Prior art keywords
- block
- embedding
- image
- watermark
- piece
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a method for embedding and extracting a reversible watermark of a digital image. The method comprises the following steps of: blocking a carrier image, judging the block types by utilizing a statistical relation between a surrounding block and a target block, adaptively selecting capacity parameters, performing block compression sensing, embedding a watermark by utilizing integer transform, embedding the watermark into a smooth and common carrier image block subjected to compression sensing for improving the watermark embedding capacity, wherein the complex carrier image block is not processed, so that the image quality and the imperceptibility are ensured; and finally, restoring the carrier image by adopting a block compression reconstructing algorithm and the reversible integer transform.
Description
Technical field
The invention belongs to the multi-media information security field; be specifically related to a kind of embedding and extracting method of the digital picture reversible water mark algorithm based on the block adaptive compressed sensing, can be used for the digital image copyright protection of the application that the data accuracies such as medical diagnosis, military image and remote sensing image processing have relatively high expectations.
Background technology
In the actual life, the digital picture of distorting and forging will be on all many-sided negative impacts that produces such as social stability, medical treatment, electronic transaction as being used to formal media, medical diagnosis, electronic bill and court's exhibit etc.Therefore, in digital picture is used, particularly the integrality of digital picture, authenticity, credibility etc. to the very crucial application of image applications in, provide a kind of integrality of digital picture, instrument of authenticity verified just to seem very necessary.
The proposition of reversible image watermark has solved the problems referred to above.Although the research origin of reversible image watermark is in lossless image authentication (being reversible image authentication), but the information that is embedded in the digital picture might not be the authentication informations such as digital signature, other information, such as the management of the content mark of medical image and retrieving information, military map and military image with license information etc., all can use reversible Image Watermarking Technique to embed fully, and in needs, extract, nondestructively recover original image simultaneously.Reversible water mark appear at the earliest Kodak in 1999 a United States Patent (USP) " Honsinger C W; Jones P; Pabani M; et al. Lossless recovery of an original image containing embedded data[P] .US; 77102/ E-C; 1998 " in, embedding grammar embedding data in image of mould is asked in utilization, so that extracting method is when recovering embedding data, can also obtain original image fully, but the embedding capacity of the method is less and produce easily salt-pepper noise.In general, reversible water mark Algorithm Performance assessment major embodiment both ways: embedding capacity and invisibility.Prior art mainly contains following three classes:
The first kind: based on histogram translation technology.This technology has people's propositions in document " Ni Z C; Shi Y Q; Ansari N; Su W. Reversible data hiding. IEEE Transactions on Circuits and Systems for Video Technology; 2006; 16 (3): 354-362 " such as Ni, and obtain in the industry extensive attention because of its low computation complexity and higher watermarking images quality, the method is according to histogrammic zero point of carrier and peak point carrying out the histogram translation, and embeds secret information.Thereafter, utilize image wavelet coefficient to have the characteristics of class class laplacian distribution, Xuan and Wu have proposed directly to turn over the figure translation algorithm based on integer wavelet transformation in succession, and higher embedding capacity and the watermarking images quality of acquisition, see document " Xuan G; Yao Q; Yang C; Gao J; Chai P; Shi Y Q, Ni Z C. Lossless data hiding using histogram shifting method based on integer wavelets. In:Proceedings of the International Workshop on Digital Watermarking. Jeju Island, Korea:Springer, 2006. 323-332 " and document " Wu X Y. Reversible semi-fragile watermarking based on histogram shifting of integer wavelet coefficients. In:Proceedings of the Digital Ecosystems and Technologies Conference. Cairns, Australia:IEEE, 2007,501-505 ".In the recent period Wang Jun auspicious wait the pertinent literature that the scholar proposed a series of histogram technologies " Wang Junxiang; Ni Jiangqun; Pan Jinwei. a kind of high-performance reversible water mark algorithm [J] based on the histogram translation. robotization journal; 2012; 38 (1): 88-96 " document " Tai W L; Yeh C M, and Chang C C. Reversible data hiding based on histogram modification of pixel differences[J]. IEEE Trans Circuits Syst Video Technology, 2009,19 (6): 906-910 "; and document " Chung Kuo-Liang, Huang Yong-Huai, Yan Wen-Ming et al. Distortion reduction for histogram modification-based reversible data hiding[J] .Applied Mathematics Computation, 2012,218 (9): 5819-5826 "
Equations of The Second Kind: based on Image Compression.This technology is by carrying out Lossless Compression to the carrier specific region, and the carrier space embed watermark information of vacating, such algorithm generally selects in the carrier visual masking and redundant larger zone to compress, in document " Fridrich J; Goljan M; Du R. Invertible authentication. In:Proceedings of the SPIE Security and Watermarking of Multimedia Contents. San Jose; USA:SPIE; 2001; 197-208 ", Fridrich etc. come embed watermark information by the Lossless Compression to jpeg image discrete cosine transform quantization parameter least significant bit (LSB), after this, Celik etc. propose a kind of universal model based on the LSB compression, and have provided the reversible water mark scheme based on bit plane.The data redundancy of considering LSB is less, and the watermark embedding capacity of this algorithm is limited.
The 3rd class: based on the difference expansion technology.Tian has at first proposed the reversible water mark method based on difference expansion, and the method generates first the difference of current pixel according to the prediction of neighbor, then, and to the move to left expansion and embed 1bit information at the lowest order of vacating of relevant difference.See document " Tian J. Reversible data embedding using a difference expansion. IEEE Transactions on Circuits and Systems for Video Technology; 2003,13 (8): 890-896 ".Alattar adopts poor expansion method to embed 2bit(or 3bit to the difference of 3 (or 4) neighbors) watermark.See document " Alattar A M. Reversible watermark using the difference expansion of a generalized integer transform. IEEE Transactions on Image Processing; 2004,13 (8): 1147-1156 ".Thodi etc. are to the poor difference expansion method embed watermark that adopts of each pixel and its prediction, and utilize the histogram modification technology to solve the problem that the location drawing transmits, see document " Thodi D M; Rodriquez J J. Prediction-error-based reversible watermarking. In:Proceedings of the International Conference on Image Processing. Singapore; Singapore:IEEE; 2004,1549-1552 ".
Yet, for the algorithm of single telescopiny, seldom have the embedding rate to reach 1.0bpp in the existing three types of technology.For improving the embedding capacity of reversible water mark algorithm, document " Chaumont M; Puech W. A high capacity reversible watermarking scheme[J]. Visual Communications and Image Processing; ser Proc SPIE; 2009; 7257:72571H " document " Coltuc D. Improved capacity reversible watermarking[C]. Proc. IEEE ICIP; San Antonio, TX, 2007,3:249-252 " the basis on a kind of high capacity reversible water mark algorithm is proposed, when adopting " Lena " image to be carrier, its embedding rate has reached 1.7bpp.Document " Weng Shaowei. the high capacity reversible water mark research [D] of digital picture. PhD dissertation, Beijing Jiaotong University, 2009 " multiple reversible water mark algorithm is proposed to improve the watermark embedding capacity.Document " Lv Lintao; imperial or royal seal. improved large capacity reversible digital watermarking embeds algorithm [J]. computer engineering; 2011,37 (22): 107-109 " improved on this basis the reversible water mark scheme of utilizing odd even characteristic and value invariant feature and difference adjusting that Weng Shaowei proposes.But these algorithm great majority have high embedding capacity and the large characteristics of distortion are unfavorable for the practical application of reversible water mark algorithm.
Summary of the invention
The purpose of this invention is to provide a kind ofDigital picture reversible water mark embedding grammar and extracting method thereof.
The present invention is digital picture reversible water mark embedding grammar and extracting method thereof, and wherein digital picture reversible water mark embedding grammar the steps include:
(1) carrier image X is divided into N non-overlapped fritter
, x
iIt is among the x arbitrary;
(2) the gained fritter the is divided three classes set of piece: smooth block set I
1, conventional bar set I
2With complex block set I
3
(3) mode of utilizing raster scanning is to embedding the district according to from right to left, and sequential scanning from top to bottom embeds district, I
1, I
2Carry out piecemeal CS accidental projection, I
3Do not do variation, correspondingly obtain the accidental projection measured value I after the compressed sensing
1 ', I
2 'Set; Utilize the self-adaptation integer transform method to determine the capacity parameter k of relevant block
x
(4) set up bit sequence record Location Map (LM) according to step (3), wherein:
Adopt simultaneously run-length encoding (Run-length encoding, RLE) that LM is carried out Lossless Compression and obtain CLM;
(5) respectively to I
1 ', I
2 'Carry out the self-adaptation integer transform, and according to k
xTo I
1 ', I
2 'Utilize self-adaptation integer transform embed watermark information, the y that record obtains
iLSB, obtained a binary sequence, utilize LSB substitute to embed CLM, binary sequence is embedded into x with integer transform
iIn, at last in conjunction with I
3Can obtain containing watermarking images I
w
Digital picture reversible water mark extracting method the steps include:
(1) with watermarking images I
wBe divided into N non-overlapped fritter
, i
WiI
wIn arbitrary fritter;
(2) use the order opposite with raster scan order in the telescopiny namely from left to right, carry out from top to bottom scanned image block, read i
WiLSB, the location drawing CLM that obtains compressing;
(3) decompress(ion) CLM namely gets location drawing LM, thereby gets TH1
,TH2 and T judge block type accordingly, obtain smooth I
1, conventional bar I
2With complex block I
3
(4) the complex block I to obtaining
3Do not process, utilize reconstructing method based on the piecemeal compressed sensing to smooth block I
1, conventional bar I
2Carry out piece and utilize restructing algorithm to be reconstructed, draw the accidental projection measured value after the compressed sensing;
(5) obtain the capacity parameter k that location drawing relevant information is determined each piece according to step (3)
x, to I in the step (4)
1, I
2Reconstruction result carry out the inverse transformation of self-adaptation integer and can obtain watermark information w and initial carrier I.
The present invention has the following advantages:
(1) the present invention is at first to carrier original image piecemeal, carry out compressed sensing according to block type, and piece is classified, adaptively selected embedding capacity coefficient carries out integer transform embedding information in each dissimilar pieces, improving watermark capacity, so that watermark capacity is large and be easy to control;
(2) the present invention is owing to carry out piecemeal compressed sensing accidental projection to smooth block and conventional bar, so that the carrier image Efficient Compression, thus effective embed watermark information;
(3) the present invention is improved so that contain the visual quality of watermarking images owing to adopted self-adaptation integer transform algorithm embed watermark information, has lowered simultaneously algorithm complex, has reduced working time;
(4) the present invention is owing to improved the watermark embedding capacity, improved and contained the watermarking images quality, the larger distortion of having avoided the noise piece to cause, more watermark is embedded into improved the watermark embedding capacity in smooth and the conventional bar and guaranteed simultaneously picture quality, invisibility is better, and then has improved the combination property of reversible water mark embedding with extracting method.
Description of drawings
Fig. 1 is object block and adjacent block distribution schematic diagram, Fig. 2 is block type partition process synoptic diagram, Fig. 3 is the watermark extracting algorithm flow chart, Fig. 4 is the performance chart that the Lena image carries out different piecemeals, Fig. 5 is the embedding rate that different carriers carries out 4 kinds of piecemeals, Fig. 6 ~ Fig. 9 is the combination property contrast of the present invention and existing algorithm, wherein Fig. 6 is Lena image comparison result, Fig. 7 Baboon image comparison result, Fig. 8 is Plane (F-16) image comparison result, and Fig. 9 is Sailboat image comparison result.
Embodiment
The present invention is digital picture reversible water mark embedding grammar and extracting method thereof, and wherein, digital picture reversible water mark embedding grammar the steps include:
(1) carrier image X is divided into N non-overlapped fritter
, x
iIt is among the x arbitrary;
(2) the gained fritter the is divided three classes set of piece: smooth block set I
1, conventional bar set I
2With complex block set I
3
(3) mode of utilizing raster scanning is to embedding the district according to from right to left, and sequential scanning from top to bottom embeds district, I
1, I
2Carry out piecemeal CS accidental projection, I
3Do not do variation, correspondingly obtain the accidental projection measured value I after the compressed sensing
1 ', I
2 'Set; Utilize the self-adaptation integer transform method to determine the capacity parameter k of relevant block
x
(4) set up bit sequence record Location Map (LM) according to step (3), wherein:
Adopt simultaneously run-length encoding (Run-length encoding, RLE) that LM is carried out Lossless Compression and obtain CLM;
(5) respectively to I
1 ', I
2 'Carry out the self-adaptation integer transform, and according to k
xTo I
1 ', I
2 'Utilize self-adaptation integer transform embed watermark information, the y that record obtains
iLSB, obtained a binary sequence, utilize LSB substitute to embed CLM, binary sequence is embedded into x with integer transform
iIn, at last in conjunction with I
3Can obtain containing watermarking images I
w
Digital picture reversible water mark extracting method the steps include:
(1) with watermarking images I
wBe divided into N non-overlapped fritter
, i
WiI
wIn arbitrary fritter;
(2) use the order opposite with raster scan order in the telescopiny namely from left to right, carry out from top to bottom scanned image block, read i
WiLSB, the location drawing CLM that obtains compressing;
(3) decompress(ion) CLM namely gets location drawing LM, thereby gets TH1
,TH2 and T judge block type accordingly, obtain smooth I
1, conventional bar I
2With complex block I
3
(4) the complex block I to obtaining
3Do not process, utilize reconstructing method based on the piecemeal compressed sensing to smooth block I
1, conventional bar I
2Carry out piece and utilize restructing algorithm to be reconstructed, draw the accidental projection measured value after the compressed sensing;
(5) obtain the capacity parameter k that location drawing relevant information is determined each piece according to step (3)
x, to I in the step (4)
1, I
2Reconstruction result carry out the inverse transformation of self-adaptation integer and can obtain watermark information w and initial carrier I.
Below in conjunction with the drawings and specific embodiments, the present invention is described in further detail.
Digital picture reversible water mark embedding grammar of the present invention, implementation step is as follows:
Step 1:Carrier image X is divided into N non-overlapped fritter
, x
iIt is among the x arbitrary;
Step 2:Calculate variance and the average of each image block, default two threshold value TH1, TH2 (TH1TH2), image block distributes as shown in Figure 1, then by the variance of piece on every side object block is judged that it is smooth block, conventional bar and complex block according to following determination methods.Whole process as shown in Figure 2, its decision method is as follows:
Case 1:((V1
TH2) ﹠amp; ﹠amp; () ﹠amp V2TH2; ﹠amp; (V3
TH2) ﹠amp; ﹠amp; (V4
TH2)): the Block0(object block) be judged to smooth block.Wherein V1, V2, V3 and V4 represent respectively the variance yields of Block1, Block2, Block3 and Block4.
Have one among Case 2:(V1, V2, V3 and the V4 at least greater than TH2, and less than TH1): Block0 is judged to conventional bar.
Case 3:((V1
TH1) ﹠amp; ﹠amp; (V2
TH1) ﹠amp; ﹠amp; (V3
TH1) ﹠amp; ﹠amp; (V4
TH1)): Block0 is judged to complex block.
Step 3:To embedding sector scanning, from left to right, order is from down to up carried out raster scanning; Then to I
1 ,I
2Piecemeal CS accidental projection adopts Orthogonal Wavelets to carry out rarefaction, and Gauss's matrix is as measuring matrix, I
3Do not do variation, can correspondingly obtain like this accidental projection measured value I after the compressed sensing
1 ', I
2 'Set; Utilize at last following method to determine the capacity parameter k of relevant block
x
For any piece x
B, the present invention will select capacity parameter k adaptively
x, generally being taken as 2 integer power, T is preset value, carries out as follows self-adaptation and determines capacity parameter:
for pixel block x sized n+1;
The variance V (x) of calculating pixel piece x;
case1: V(x) > T,
do k
x= 1;
case2: T/9 < V(x) <= T ,
Label 1: judge whether x belongs to B2;
if(x
B2)，then
k
x = 2;
else
k
x = 1;
end if;
case 3: T/49 < V(x) <= T/9;
Label 2: judge whether x belongs to B4;
if(x
B4），then
k
x = 4;
else
go to label 1;
end if
case 4: V(x) <= T/49
Judge whether x belongs to B8;
if(x
B8），then
k
x = 8;
else
go to label 2;
end if;
end
Step 4:Set up bit sequence record Location Map (LM) according to step 3, wherein:
Adopt simultaneously run-length encoding (Run-length encoding, RLE) that LM is carried out Lossless Compression and obtain CLM.
Step 5:Select first a positive integer as embedding number of levels L, utilize pseudo random function rand (1, m) generate the binary pseudo-random W that length is m, with this sequence as watermark sequence, wherein m represents the length of watermark sequence W, be the dimension of watermark in the watermark sequence, m is less than or equal to the number of piece in the embedded block sequence; Again respectively to I
1 ', I
2 'Carry out the self-adaptation integer transform, and according to k
xTo I
1 ', I
2 'Utilize self-adaptation integer transform embed watermark information, self-adaptation integer transform process is as follows:
The y that record obtains
iLSB, obtained a binary sequence, utilize LSB substitute to embed CLM, binary sequence is embedded into x with integer transform
iIn, at last in conjunction with I
3Can obtain containing watermarking images I
w
By above-mentioned
Step 1-
Step 5Can realize the embedding of watermark, the pseudorandom watermark sequence W of two-value is embedded into I in the original image, obtain containing watermarking images I
w
As shown in Figure 3, the present invention is based on the digital picture reversible water mark extracting method of block adaptive compressed sensing, implementation step is as follows:
Step 1:With watermarking images I
wBe divided into N non-overlapped fritter
, i
WiI
wIn arbitrary fritter;
Step 2:With the order opposite with raster scan order in the telescopiny namely from left to right, carry out from top to bottom scanned image block, read i
WiLSB, the location drawing CLM that obtains compressing;
Step 3:The decompress(ion) CLM that extracts is namely got location drawing LM, thereby get TH1
,TH2 and T judge block type accordingly, obtain smooth I
1, conventional bar I
2With complex block I
3
Step 4:At first use reconstructing method based on the piece compressed sensing to smooth block I
1, conventional bar I
2Carry out piece and utilize restructing algorithm to be reconstructed, complex block I
3Do not process, restructuring procedure carries out under Orthogonal Wavelets, and Gauss's matrix adopts Nonlinear conjugate gradient algorithm and band to recall the reconstruct that line search is optimized target for measuring matrix, draws at last the accidental projection measured value after the compressed sensing;
Step 5:Obtain the capacity parameter k that location drawing relevant information is determined each piece according to step 3
x, to I in the step 4
1, I
2Reconstruction result carry out the inverse transformation of self-adaptation integer and can obtain watermark information w and initial carrier I.
Side's of the present invention advantage can further specify by following emulation experiment:
1, experiment condition and description of test:
All test the present invention all at Dell Inspiron N4010 notebook computer, Intel (R) Core (IM) i3 CPU, and the 2G internal memory obtains take MATLAB7.8 as experiment porch under the Windows7 operating system.Carrier image comprises Lena, Baboon, Plane (F-16) and Sailboat, and the random bit sequence that employing random function (rand) generates at random is as watermark information.Performance evaluation criterion embeds behind bit rate and the embedding data visual quality of image with data and weighs, the former weighs by the ratio (BPP) of computational data embedding capacity and image size, the latter judges this key property of invisibility of watermark by Y-PSNR (PSNR, Peak signal-noise ratio).
2, experiment content:
Experiment one:Different piecemeal numbers are on the impact of algorithm performance
The embedding rate of each piece of the present invention is
, embedding capacity increases along with the increase of tile size.But if block is too large, and its noise is larger, and variance is also larger.Therefore, balance embedding capacity and degree of distortion, it is very important selecting suitable piece.
Fig. 4 reflects the performance curve of different piecemeals take Lena as example, can find out that embedding capacity was higher and the invisible performance of image is better by piecemeal 2 * 2 to 4 * 4 o'clock.Yet performance occupies between the two when being increased to 8 * 8, and the PSNR value is a little more than 4 * 4 when piecemeal is 16 * 16, and maximum embedding capacity reaches 1.8bpp, and the maximum embedding capacity of other piecemeals all is lower than this value.Its main cause is that on the one hand the larger embedding rate of piece is just larger, but the point that the too large words of piecemeal can not embed is many, thereby location map is larger; Owing to adopted the piecemeal compressive sensing theory in this algorithm, different block sizes can affect to be rebuild effect and time on the other hand.Less along with block size, the obvious block boundary of the easier formation in border of subspace in reconstructed image (or piece), visual effect reduces gradually, the Y-PSNR that draws also reduces gradually, when piecemeal is 16 * 16 its rebuild effect and better and reconstruction time moderate, so show optimal performance, and 8 * 8 piecemeals when rebuilding blocking effect relatively outstanding, affected its reconstruction quality.
The present invention has adopted 4 width of cloth carrier images to carry out respectively 2 * 2,4 * 4,8 * 8 and 16 * 16 piecemeals and has embedded, the maximum embedding rate histogram of its correspondence as shown in Figure 5, through many experiments, getting best piecemeal is 16 * 16 piecemeals.Because 4 width of cloth carrier texture smoothnesses are different, maximum embedding rate is also variant after adopting embedding grammar of the present invention, and wherein the maximum embedding rate of Lena reaches 1.8bpp, and the maximum embedding rate of Baboon is 0.8bpp, the maximum embedding rate of Plane is 1.87bpp, and the maximum embedding rate of Sailboat is 1.6bpp.
Experiment two:Algorithm synthesis performance and other literature methods are relatively
In order to verify the superiority of BACS-RWA algorithm of the present invention, experimental selection 16 * 16 piecemeals, adopt same vehicle image and watermark, choose Tsai P etc. document " Tsai P; Hu Y C; and Yeh H L. Reversible image hiding scheme using predictive coding and histogram shifting[J]. Signal Processing; 2009; 89 (6): 1129-1143 " (being called for short " Tsai P method "), the document of Luo Li-xin " Luo Li-xin; Zheng yong; Chen Ming; et al. Reversible image watermarking using interpolation technique[J]. IEEE Transactions on Information Forensics and Security; 2010; 5 (1): 187-193 " (be called for short " Luo Li-xin method ") and Luo Jiangao document " Luo Jiangao. reversible image watermark and reversible graph are as Research of Authentication Techniques [D]. PhD dissertation, South China Science ﹠ Engineering University, 2011 " three kinds of reversible water mark algorithms compare experiment (to be called for short " Luo Jian-gao method ").Can find out that from Fig. 6 ~ Fig. 9 combination property contrast and experiment BACS-RWA algorithm of the present invention has preferably watermark capacity-distortion performance, performance curve is more smooth simultaneously, and algorithm changes more steady.
Fig. 6 is based on the combination property correlation curve figure that localized watermark embeds the less Lena image of ability difference.Quality under height embedding rate is higher, and its PSNR value still is higher than 35dB during up to 1bpp in the embedding rate; When the embedding rate less than 0.4 the time, Luo Jian-gao method is not too obvious in the effect that the piecemeal processing list reveals, the PSNR performance is a little less than Luo Li-xin method, and Tsai P method performance all is lower than other algorithms in the 1bpp scope, and this is to be the minimum scheme that merges compression that merges of grey level histogram because its gray scale overflows employing.The highest embedding capacity of algorithm of the present invention can reach 1.8bpp, the not enough 1.4bpp of Luo Jian-gao method embedding capacity, the not enough 1.2bpp of Luo Li-xin method embedding capacity, and the maximum embedding capacity of Tsai P method 1bpp only.
Fig. 7 is based on the combination property correlation curve that carrier image is Baboon.Because Baboon is the most complicated image of texture in 4 width of cloth images, the quality of image is along with increasing of embedding data amount descends very fast, 4 kinds of method PSNR value differences are different not obvious when the embedding rate is 0.1bpp, when the embedding rate reaches 0.4bpp, the PSNR that utilizes the image behind the BACS-RWA embedding data has shown good performance than Luo Jian-gao method, Luo Li-xin method and Tsai P method respectively high 2dB, 4dB, 5dB.Along with embedding increasing of quantity of information, complicated texture is more and more serious on the picture quality impact, but its PSNR still is higher than other algorithms.
It is carrier image that Fig. 8 has adopted the comparatively simple Plane (F-16) of texture, integral image is the most level and smooth with respect to other 3 kinds of carriers, the difference of block of pixels is little, uses less threshold value then image can be divided into different types, and smooth region can produce more embedding Information Availability pixel.Therefore, the embedding rate is higher, and picture quality is better, and combination property is more stable.And near the PSNR of other three kinds of algorithms image when the embedding rate reaches the 1.2bpp has all reached 30dB, and in the time of near the embedding rate is 0.7bpp, Luo Jian-gao method and algorithm performance of the present invention differ less, and the maximum embedding rate of the present invention reaches maximal value 1.87bpp.
Fig. 9 tests comparing result to Sailboat, the embedding rate near 0.2bpp the time Luo Li-xin method PSNR value be slightly higher than algorithm of the present invention, but the increase along with its embedding rate, the PSNR value of Luo Li-xin method and Tsai P method obviously descends, and Luo Jian-gao method is not tested this carrier figure, combination property curve of the present invention is then more steady, and its PSNR value still is higher than 24dB when reaching maximum embedding rate.
The present invention chooses three fixedly embedded rate 0.5bpp, 1.0bpp and 1.2bpp carries out corresponding PSNR contrast, result such as table 1.
PSNR value contrast under the identical embedding rate of table 1
Wherein the unit of PSNR is dB, and "---" expression in the table 1 is because the image self character can't carry out corresponding experiment measuring.As can be seen from the table, in the identical situation of selected embedding rate:
When the embedding rate is 0.5bpp, mean P SNR of the present invention is higher than Luo Li-xin method 2.84dB, is higher than Luo Jian-gao method 2.1dB; Along with the embedding rate raises;
When reaching 1.0bpp, mean P SNR of the present invention is higher than document Luo Li-xin method 2.3dB, is higher than Luo Jian-gao method 1.3dB;
When identical embedding rate was 1.2bpp, mean P SNR of the present invention was higher than respectively Luo Li-xin method 3.15dB, is higher than Luo Jian-gao method 1.45dB.
This difference main cause occurs and be linear prediction error that Tsai P method adopts and compare the predicated error of other algorithms and will disperse, what gray scale was overflowed employing is the minimum scheme that merges compression that merges of grey level histogram.Comparatively speaking, this algorithm performance is slightly poor, but Performance Ratio is more steady, and this gives the credit to its grey level histogram adjustment and has adopted the minimum scheme that merges item.Luo Li-xin method adopts the reversible image watermark based on the interpolation technique of histogram displacement, owing to obtain the predicated error histogram of concentrating, algorithm performance is higher when little embedded quantity, but not enough is that gray scale overflow control scheme is for the host image that contains the larger proportion pixel at bright or the darkest end, because supplementary can't complete preservation cause embedding inefficacy.Luo Jian-gao method embeds many that the strong regional area watermark of ability distributes in watermark, embeds lacking of weak distribution, thereby but its inevitably produced distortion after by iterative processing and affected picture quality.This shows that block adaptive compressed sensing of the present invention embeds algorithm and has preferably picture quality, be suitable for the high power capacity watermark information and embed.
In sum, the larger distortion that the present invention has avoided the noise piece to cause, more watermark is embedded into improved the watermark embedding capacity in smooth and the conventional bar and guaranteed simultaneously picture quality, invisibility is better, and the algorithm computing is simple, and the reversible water mark algorithm was compared and all have superiority on the combination properties such as embedding capacity and invisibility more in the past.
Claims (4)
1. digital picture reversible water mark embedding grammar the steps include:
(1) carrier image X is divided into N non-overlapped fritter
, x
iIt is among the x arbitrary;
(2) the gained fritter the is divided three classes set of piece: smooth block set I
1, conventional bar set I
2With complex block set I
3
(3) mode of utilizing raster scanning is to embedding the district according to from right to left, and sequential scanning from top to bottom embeds district, I
1, I
2Carry out piecemeal CS accidental projection, I
3Do not do variation, correspondingly obtain the accidental projection measured value I after the compressed sensing
1 ', I
2 'Set; Utilize the self-adaptation integer transform method to determine the capacity parameter k of relevant block
x
(4) set up bit sequence record Location Map (LM) according to step (3), wherein:
Adopt simultaneously run-length encoding (Run-length encoding, RLE) that LM is carried out Lossless Compression and obtain CLM;
(5) respectively to I
1 ', I
2 'Carry out the self-adaptation integer transform, and according to k
xTo I
1 ', I
2 'Utilize self-adaptation integer transform embed watermark information, the y that record obtains
iLSB, obtained a binary sequence, utilize LSB substitute to embed CLM, binary sequence is embedded into x with integer transform
iIn, at last in conjunction with I
3Can obtain containing watermarking images I
w
2.
According toDigital picture reversible water mark embedding grammar claimed in claim 1 is characterized in that the classification of three class set of blocks, the steps include:
(1) carrier image is carried out piecemeal, and calculate variance and the average of each image block, by the relation of image block difference around target image piece and its and predetermined threshold value, judge different images piece texture complexity, thereby the embedding different pieces of information, default two threshold value TH1, TH2, TH1
TH2;
(2) by the variance of piece on every side object block is judged that it is smooth block, conventional bar, complex block.
According to claim 1 with 2 described digital picture reversible water mark embedding grammars, it is characterized in that according to embedding the district from right to left, sequential scanning from top to bottom embeds the district, and the embedded block that satisfies condition is accordingly carried out piecemeal CS accidental projection, obtains the accidental projection measured value.
4. digital picture reversible water mark extracting method the steps include:
(1) with watermarking images I
wBe divided into N non-overlapped fritter
, i
WiI
wIn arbitrary fritter;
(2) use the order opposite with raster scan order in the telescopiny namely from left to right, carry out from top to bottom scanned image block, read i
WiLSB, the location drawing CLM that obtains compressing;
(3) decompress(ion) CLM namely gets location drawing LM, thereby gets TH1
,TH2 and T judge block type accordingly, obtain smooth I
1, conventional bar I
2With complex block I
3
(4) the complex block I to obtaining
3Do not process, utilize reconstructing method based on the piecemeal compressed sensing to smooth block I
1, conventional bar I
2Carry out piece and utilize restructing algorithm to be reconstructed, draw the accidental projection measured value after the compressed sensing;
(5) obtain the capacity parameter k that location drawing relevant information is determined each piece according to step (3)
x, to I in the step (4)
1, I
2Reconstruction result carry out the inverse transformation of self-adaptation integer and can obtain watermark information w and initial carrier I.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012104094507A CN102903076A (en) | 2012-10-24 | 2012-10-24 | Method for embedding and extracting reversible watermark of digital image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012104094507A CN102903076A (en) | 2012-10-24 | 2012-10-24 | Method for embedding and extracting reversible watermark of digital image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102903076A true CN102903076A (en) | 2013-01-30 |
Family
ID=47575290
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012104094507A Pending CN102903076A (en) | 2012-10-24 | 2012-10-24 | Method for embedding and extracting reversible watermark of digital image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102903076A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103577730A (en) * | 2013-11-15 | 2014-02-12 | 桂林理工大学 | Reversible database watermark embedding and extracting method based on integral wavelet transformation |
| CN103646376A (en) * | 2013-12-25 | 2014-03-19 | 北京德信易税网络技术有限公司 | Digital watermark image generation method |
| CN103700061A (en) * | 2013-12-30 | 2014-04-02 | 东北大学 | Compressive-sensing-based digital image watermark embedding and extraction method |
| CN104284190A (en) * | 2014-11-05 | 2015-01-14 | 安徽大学 | Compressed image steganography encoding method based on AMBTC high and low average optimization |
| CN104751402A (en) * | 2015-04-16 | 2015-07-01 | 西安电子科技大学 | Method for embedding robust and reversible watermark |
| CN105488822A (en) * | 2015-12-15 | 2016-04-13 | 西华大学 | Reversible image hiding algorithm based on AMBTC algorithm |
| CN105979268A (en) * | 2016-05-05 | 2016-09-28 | 北京智捷伟讯科技有限公司 | Safe information transmission method based on lossless watermark embedding and safe video hiding |
| CN106485642A (en) * | 2016-09-30 | 2017-03-08 | 北京交通大学 | The method of embedded visible watermark in three-dimensional grid model |
| CN106709853A (en) * | 2016-11-30 | 2017-05-24 | 开易（北京）科技有限公司 | Image retrieval method and system |
| CN108022197A (en) * | 2017-11-21 | 2018-05-11 | 安徽大学 | A kind of medical image reversible information hidden method based on the division of multi-texturing degree |
| CN111062853A (en) * | 2019-12-20 | 2020-04-24 | 中国科学院自动化研究所 | Self-adaptive image watermark embedding method and system and self-adaptive image watermark extracting method and system |
| CN111277831A (en) * | 2020-01-15 | 2020-06-12 | 广东工业大学 | Image compression method and system based on classified coding |
| CN112907432A (en) * | 2021-03-08 | 2021-06-04 | 桂林理工大学 | Reversible watermark data hiding method based on bilinear interpolation |
| CN113538198A (en) * | 2020-04-15 | 2021-10-22 | 北京达佳互联信息技术有限公司 | Watermark adding method, device, storage medium and electronic equipment |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101034468A (en) * | 2007-04-05 | 2007-09-12 | 上海交通大学 | Lossless digital watermark method having regioselectivity |
| US20100177977A1 (en) * | 2009-01-15 | 2010-07-15 | Google Inc. | Image Watermarking |
| CN101833746A (en) * | 2010-04-23 | 2010-09-15 | 北京大学 | Method for embedding reversible watermark in digital image and method for extracting reversible watermark from digital image |
| CN102044061A (en) * | 2010-12-20 | 2011-05-04 | 北京大学 | Embedding method and extracting method of reversible watermarks |
| CN102147912A (en) * | 2011-03-30 | 2011-08-10 | 北京航空航天大学 | Adaptive difference expansion-based reversible image watermarking method |
| CN102194204A (en) * | 2010-03-15 | 2011-09-21 | 北京大学 | Method and device for embedding and extracting reversible watermarking as well as method and device for recovering image |
-
2012
- 2012-10-24 CN CN2012104094507A patent/CN102903076A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101034468A (en) * | 2007-04-05 | 2007-09-12 | 上海交通大学 | Lossless digital watermark method having regioselectivity |
| US20100177977A1 (en) * | 2009-01-15 | 2010-07-15 | Google Inc. | Image Watermarking |
| CN102194204A (en) * | 2010-03-15 | 2011-09-21 | 北京大学 | Method and device for embedding and extracting reversible watermarking as well as method and device for recovering image |
| CN101833746A (en) * | 2010-04-23 | 2010-09-15 | 北京大学 | Method for embedding reversible watermark in digital image and method for extracting reversible watermark from digital image |
| CN102044061A (en) * | 2010-12-20 | 2011-05-04 | 北京大学 | Embedding method and extracting method of reversible watermarks |
| CN102147912A (en) * | 2011-03-30 | 2011-08-10 | 北京航空航天大学 | Adaptive difference expansion-based reversible image watermarking method |
Non-Patent Citations (2)
| Title |
|---|
| 张俊兰等: "基于小波变换的图象认证的数字水印技术", 《微型电脑应用》 * |
| 罗剑高等: "基于自适应图像块组合的无损图像认证算法", 《通信学报》 * |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103577730A (en) * | 2013-11-15 | 2014-02-12 | 桂林理工大学 | Reversible database watermark embedding and extracting method based on integral wavelet transformation |
| CN103646376A (en) * | 2013-12-25 | 2014-03-19 | 北京德信易税网络技术有限公司 | Digital watermark image generation method |
| CN103700061B (en) * | 2013-12-30 | 2016-08-17 | 东北大学 | A kind of digital figure watermark based on compressed sensing embeds and extracting method |
| CN103700061A (en) * | 2013-12-30 | 2014-04-02 | 东北大学 | Compressive-sensing-based digital image watermark embedding and extraction method |
| CN104284190A (en) * | 2014-11-05 | 2015-01-14 | 安徽大学 | Compressed image steganography encoding method based on AMBTC high and low average optimization |
| CN104284190B (en) * | 2014-11-05 | 2017-05-10 | 安徽大学 | Compressed image steganography encoding method based on AMBTC high and low average optimization |
| CN104751402A (en) * | 2015-04-16 | 2015-07-01 | 西安电子科技大学 | Method for embedding robust and reversible watermark |
| CN104751402B (en) * | 2015-04-16 | 2018-07-27 | 西安电子科技大学 | A kind of method of robust reversible watermark insertion |
| CN105488822A (en) * | 2015-12-15 | 2016-04-13 | 西华大学 | Reversible image hiding algorithm based on AMBTC algorithm |
| CN105488822B (en) * | 2015-12-15 | 2019-06-21 | 西华大学 | Reversible image hidden algorithm based on AMBTC algorithm |
| CN105979268A (en) * | 2016-05-05 | 2016-09-28 | 北京智捷伟讯科技有限公司 | Safe information transmission method based on lossless watermark embedding and safe video hiding |
| CN106485642A (en) * | 2016-09-30 | 2017-03-08 | 北京交通大学 | The method of embedded visible watermark in three-dimensional grid model |
| CN106485642B (en) * | 2016-09-30 | 2019-05-24 | 北京交通大学 | The method of visible watermark is embedded in three-dimensional grid model |
| CN106709853A (en) * | 2016-11-30 | 2017-05-24 | 开易（北京）科技有限公司 | Image retrieval method and system |
| CN108022197A (en) * | 2017-11-21 | 2018-05-11 | 安徽大学 | A kind of medical image reversible information hidden method based on the division of multi-texturing degree |
| CN108022197B (en) * | 2017-11-21 | 2021-03-12 | 安徽大学 | Reversible information hiding method for medical image based on multi-texture degree division |
| CN111062853A (en) * | 2019-12-20 | 2020-04-24 | 中国科学院自动化研究所 | Self-adaptive image watermark embedding method and system and self-adaptive image watermark extracting method and system |
| CN111277831A (en) * | 2020-01-15 | 2020-06-12 | 广东工业大学 | Image compression method and system based on classified coding |
| CN111277831B (en) * | 2020-01-15 | 2022-05-06 | 广东工业大学 | Image compression method and system based on classified coding |
| CN113538198A (en) * | 2020-04-15 | 2021-10-22 | 北京达佳互联信息技术有限公司 | Watermark adding method, device, storage medium and electronic equipment |
| CN113538198B (en) * | 2020-04-15 | 2023-02-28 | 北京达佳互联信息技术有限公司 | Watermark adding method, device, storage medium and electronic equipment |
| CN112907432A (en) * | 2021-03-08 | 2021-06-04 | 桂林理工大学 | Reversible watermark data hiding method based on bilinear interpolation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102903076A (en) | Method for embedding and extracting reversible watermark of digital image | |
| Khan et al. | A recent survey of reversible watermarking techniques | |
| Peng et al. | Adaptive reversible data hiding scheme based on integer transform | |
| Wu et al. | Reversible image watermarking on prediction errors by efficient histogram modification | |
| Qin et al. | An inpainting-assisted reversible steganographic scheme using a histogram shifting mechanism | |
| Abdelwahab et al. | A discrete wavelet transform based technique for image data hiding | |
| Qin et al. | Visible watermark removal scheme based on reversible data hiding and image inpainting | |
| Liu et al. | Reducing location map in prediction-based difference expansion for reversible image data embedding | |
| CN101105857B (en) | High capacity reversible water mark method based on predication and companding technology | |
| CN110232651B (en) | Reversible information hiding method based on block prediction and differential value histogram translation | |
| Al-Mansoori et al. | Robust watermarking technique based on DCT to protect the ownership of DubaiSat-1 images against attacks | |
| Peng et al. | Reversible data hiding based on RSBEMD coding and adaptive multi-segment left and right histogram shifting | |
| Wu et al. | A high capacity reversible data hiding scheme with edge prediction and difference expansion | |
| Yang et al. | Efficient reversible data hiding algorithm based on gradient-based edge direction prediction | |
| Joshi et al. | Real time implementation of digital watermarking algorithm for image and video application | |
| Gao et al. | A new reversible watermarking scheme based on Integer DCT for medical images | |
| KR101026081B1 (en) | Reversible watermark inserting and original image restoring methods | |
| CN111065000B (en) | Video watermark processing method, device and storage medium | |
| Ghosh et al. | FPGA and SoC based VLSI architecture of reversible watermarking using rhombus interpolation by difference expansion | |
| Kunhu et al. | A reversible watermarking scheme for ownership protection and authentication of medical images | |
| Wu et al. | Reversible data hiding based on medical image systems by means of histogram strategy | |
| Mathews et al. | Histogram shifting based reversible data hiding using block division and pixel differences | |
| Weng et al. | Adaptive reversible data hiding based on a local smoothness estimator | |
| Luo et al. | Reversible image water marking based on prediction-error expansion and compensation | |
| CN110570343B (en) | Image watermark embedding method and device based on self-adaptive feature point extraction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20130130 |