CN1540599A - Vulnerable watermark method for protecting electronic signet - Google Patents
Vulnerable watermark method for protecting electronic signet Download PDFInfo
- Publication number
- CN1540599A CN1540599A CNA2003101119289A CN200310111928A CN1540599A CN 1540599 A CN1540599 A CN 1540599A CN A2003101119289 A CNA2003101119289 A CN A2003101119289A CN 200310111928 A CN200310111928 A CN 200310111928A CN 1540599 A CN1540599 A CN 1540599A
- Authority
- CN
- China
- Prior art keywords
- watermark
- wavelet
- original
- image
- interpolation
- 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
Images
Abstract
Treating original watermark in binary form, the invention constructs pyramidal watermark with descended resolution layer by layer. Meanwhile, using wavelet transform decomposes multimedia data. Then, according to certain rule, some special wavelet coefficients are selected to embed to watermark. Finally, multimedia data with watermark being embedded into can be obtained through wavelet inverse transform. The invention possesses capability of detecting interpolation including interpolation characteristic and region. The invention also discloses rules of classifying interpolation to distinguish occasional interpolation from vicious interpolation. Thus, vulnerable watermark can resist such as JPEG loss compression etc. content lockout operation. The invention insures reliability of multimedia data released or transferred on network so that it is applicable to authentication possibly.
Description
Technical field
The present invention relates to a kind of fragile watermark technology, belong to field of multimedia signal processing based on wavelet transformation.
Background technology
Fragile watermark is to any variation sensitivity that adds all in image.Whether, truly whether and whether complete existence by detecting embed watermark can guarantee the confidence level of original image.In general, fragile watermark need satisfy following three basic demands: the 1) high susceptibility to distorting; 2) invisibility; 3) security.In recent years, because traditional multimedia messages authentication techniques: it is more and more undesirable that digital signature (Digital Signature) is shown when carrying out multimedia authentication, and fragile watermark becomes the new research focus of authentication techniques gradually.Fragile watermark provides new approaches for the authenticity and integrity authentication of multimedia messages.As effective means of supplementing out economy of digital signature, fragile watermark has broad application prospects.
The research of fragile watermark concentrates on the image watermark field at present.Its technology can be divided into two classes: spatial domain technology and transform domain technology.The former comes embed watermark by direct modification image pixel value.The latter then adopts the superperformance of image transformation, chooses special conversion coefficient and carries out the watermark embedding.Because wavelet transform DWT (Discrete Wavelet Transform) has when good-frequency localization property and the transformation mechanism that conforms to human visual system HVS (Human Visual System), thereby in Joint Photographic Experts Group JPEG2000 of new generation, occupied core position, and replaced discrete cosine transform DCT (Discrete Cosine Transform) to become the main tool of transform domain digital watermarking algorithm gradually.
Existing Fragile Watermark Algorithm is a lot, but its watermark performance and algorithm security can't be accomplished to take into account.Especially based on the fragile watermark scheme of DWT, because its research is at the early-stage, watermark is only in performance, and all needs to be improved significantly in security.In addition, the fragile watermark embedding formula in conjunction with HVS does not still have comparatively successful precedent with the research of distorting classifying rules yet.
Summary of the invention
The objective of the invention is to propose a kind of fragile watermark embedding grammar, be used for that the E-seal image is carried out easily multiresolution and distort and detect and distort classification, finish the authenticity and integrity authentication, improve the confidence level of official document based on wavelet transformation.
To achieve these goals, the inventive method is made up of following four steps:
1) the pyramidal structure of watermark; 2) in conjunction with human visual system's embed watermark; 3) watermark extracting; 4) multiresolution is distorted and is detected and authentication.
Structure original watermark pyramid is the method for successively decreasing according to the resolution that the JBIG of bianry image expert group (Joint Bi-level Image expert Group) Standard of image compression provides, can the branch many levels construct, the present invention divides three levels to construct.
Watermark in conjunction with human visual system HVS embeds, the quantization threshold matrix that adopts people such as Watson in the table 2 to propose, original image is done quantification modulation to greatest extent in wavelet transformed domain, so that make any visually visible content change and distort all and will be detected.
Multiresolution is distorted to detect and is based on the watermark pyramid and in conjunction with the watermark embedded technology of human visual system HVS, all doing watermark under each level of resolution embeds, thereby make that distorting characteristic all obtains embodying, and finishes multiresolution and distorts detection under each level of resolution; Multiresolution is distorted detection and is undertaken by distorting classifying rules, according to calculating the difference figure D that extracts between watermark and the original watermark
l(i, j)=| W
l(i, j)-W '
l(i, j) | judge:
(1) for any l, λ
l=0, there is not modification so; (2), make λ if there is l
l>0 and, δ
1<α thinks accidental modification so; (3) for any l, if δ
l〉=α thinks malicious attack.Wherein l is a level of resolution, λ
l, δ
lBe respectively that the detection erroneous point of distorting of watermark embedding subband is densely distributed ratio shared in this subband is had a few with distorting to detect in the erroneous point under the level of resolution l.α is the real number between predefined 0.5 to 1, generally is taken as 0.55.
Compare with existing fragile watermark technology, the present invention has following advantage:
1) in conjunction with the human visual system, carry out watermark and embed, guarantee the invisibility of watermark.
2) multiresolution is distorted detection, guarantees to distort the integrality of characteristic, and is rich.
3) distort the proposition of classifying rules, distinguish accidental the modification and malicious attack.
4) compression of anti-JPEG (joint photographic experts group) JPEG (Joint Photogragh Coding Experts Group) compression standard.
Below we illustrate the good effect that the present invention brings from theoretical and experimental data.
1) the pyramidal structure of watermark
Among the present invention, in order adapting, and to embed easily, to reach the purpose that multiresolution is distorted detection simultaneously, must do original watermark and anticipate, construct a watermark pyramid that resolution is successively successively decreased with the coefficient distributed architecture behind the wavelet transformation.The resolution diminishing method that provides according to the JBIG of bianry image expert group (JointBi-level Image expert Group) standard in the invention calculates a watermark pyramid.Result of calculation such as Fig. 1.
2) embedding scheme
Former figure is carried out wavelet decomposition, note f
K, lFor decomposing the wavelet coefficient on the l level k direction of back.Wherein k=h (level), v (vertically), d (diagonal angle); L=1 ..., L.For embed watermark in the subband on certain specific direction of l level, need to calculate
Wherein JND () is a quantization matrix, and concrete numerical value sees Table 2,
Be floor function, f
l(i j) is (i, the wavelet coefficient of j) locating in this subband.Order
For previously selected constant m, interval [s JND (and i, j), (s+1) JND (i, j)] be divided into 2
M-1Individual minizone.According to the interval sequence number at current coefficient place, we can finish embedding by the corresponding modify coefficient, remember that this interval sequence number is t ∈ [1,2
M-1], the interval long length that is.Watermark embeds the m position at every turn, get first of current m position watermark for w (i, j).Remaining m-1 gets its decimal integer value and is designated as r in the position, and scope is [0,2
M-1].Then amended coefficient is:
Wherein
In order to strengthen the security of watermark, can carry out scramble to watermark data and embedded location, and use cipher controlled.Like this, the assailant is difficult under the situation that does not have priori and recovers watermark.Fig. 2 has provided the particular flow sheet that watermark embeds.
3) watermark extracting
The same with the watermark embedding, at first testing image is carried out wavelet decomposition, the coefficient of keeping the score after separating is f ', and is same, for the embed watermark subband, calculates
Order
Then have
So, the watermark of extraction is:
w=Q′
i,j·2
m-1+t′
Then the binary value of a demand w just can obtain the m position and extract watermark.
4) distort detection and verification process
Make W ' l (i, j) watermark for extracting.Calculate difference figure
D
l(i,j)=|W
l(i,j)-W′
l(i,j)|
In order to provide the authentication result judgment criteria of a quantification, at first need the frequency field that each watermark embeds is done as giving a definition:
Dense point: (i is dense to be meant that its eight close on picture element and have at least one to be to detect erroneous point j) to current watermark detection erroneous point x.
Sparse point: (i is sparse be meant eight to close on the picture element neither one be to detect erroneous point j) to current watermark detection erroneous point x.
Area
L, dense={ dense district area }={ dense some number };
Area
L, sparse={ rarefaction area }={ sparse some number };
area
l,total=area
l,dense+area
l,sparse;
Area
l={ l piecemeal area }={ l piecemeal picture element sum };
Be defined as follows judgment criterion again:
(1) for any l, λ
l=0, there is not modification so.
(2), make λ if there is l
l>0 and, δ
l<α thinks accidental modification so.α is a real number of getting in advance between fixed 0.5 to 1.
(3) for any l, δ
l〉=α thinks malicious attack.
Make authentication result at last, do not revise or accidental the modification if having, image then to be checked is true; If malicious attack, image then to be checked is false.The multiresolution watermark difference figure that detection obtains can obtain the comprehensive detection figure that distorts through image fusion technology.Distort detection and identifying procedure as shown in Figure 4.
In the electronic government affairs system, official document comes and goes except the ID authentication mechanism of strictness, and the appearance of E-seal also provides an extra authentication authority for authentication.E-seal truly whether, reflected directly whether official document true.Therefore, the E-seal of fragile watermark protection provides very high security and stability for electronic government affairs system.
Description of drawings
Fig. 1 is a watermark pyramid construction process.
Fig. 2 is that watermark embeds process flow diagram.
Fig. 3 be before and after the different original image embed watermark effect relatively.
Fig. 4 is watermark extracting and distorts the process flow diagram of detection.
Fig. 5 distorts detection example.
Fig. 6 is the anti-JPEG compressed capability of an algorithm experimental result.
Fig. 7 is accidental attack detecting example.
Among Fig. 1-7, the 1st, watermark pyramid ground floor, the 2nd, the watermark pyramid second layer, the 3rd, the 3rd layer on watermark pyramid, the 4th, original watermark, the 5th, original image, the 6th, three grades of wavelet decomposition figure of original image, the 7th, the high-frequency sub-band (HH of first order wavelet decomposition
1), the 8th, the high-frequency sub-band (HH of second level wavelet decomposition
2), the 9th, the low frequency sub-band (LL of third level wavelet decomposition
3), the 10th, the image behind the embed watermark (watermarking images), the 11st, the original watermark pyramid, the 12nd, extract the watermark pyramid, the 13rd, image to be detected, the 14th, multiresolution watermark difference figure, the 15th, watermark extracting figure, the 16th, synthetic watermark difference figure, the 17th, distort judgement, classifying rules, the 18th, the seal after distorting, the 19th, with distorting that partial content brightness increases.
Embodiment
Among Fig. 1, original watermark 4 is a bianry image, through being divided into 1,2,3 three levels after the resolution decrement operations.
The embodiment that Fig. 2 embeds for watermark.Among Fig. 2, original watermark 4 is a bianry image, through being divided into 1,2,3 three levels of watermark pyramid 11 after the resolution decrement operations; Original image 5 is a gray-scale map, gets wavelet coefficient 6 through three grades of wavelet decomposition, according to aforementioned embedding scheme, with 1 layer of watermark pyramid 11 embed first order wavelet decomposition to angular direction (HH
1Subband) coefficient 7, with 2 layers of watermark pyramid 11 embed second level wavelet decomposition to angular direction (HH
2Subband) coefficient 8, with 3 layers of low frequency (LL that embeds third level wavelet decomposition of watermark pyramid 11
3Subband) coefficient 9, after embedding finishes coefficient carried out inverse wavelet transform and get watermarking images 10.
Among Fig. 3, the 5th, original image, the 10th, the image behind the embed watermark (watermarking images).
Among Fig. 4, treating detected image 13 carries out three grades of wavelet decomposition and extract multiresolution watermark pyramid 12 in corresponding subband, multiresolution watermark pyramid gets watermark extracting Figure 15 on the one hand after image co-registration, on the other hand with the original watermark pyramid subtract each other multiresolution watermark difference Figure 14, whether multiresolution watermark difference figure gets watermark difference Figure 16 after image co-registration, distorted from visually reflecting image to be detected intuitively; Multiresolution watermark difference desired to make money or profit judges and classify with classifying rules 17 and can get classification results with distorting, comprise do not have distort, accidental distorting with malice distort.
Distort in Fig. 5 malice and to detect among the embodiment, to " Baboon " in the watermarking images 10 figure, the content in the black surround will be distorted, and replaces to such an extent that distort back image 13 with the content of original image 5 same positions; Extract watermark according to the step of Fig. 4 and obtain corresponding watermark extracting Figure 15 and watermark difference Figure 16, thereby can find and positioning tampering.If threshold value is 0.55, a dense as calculated distribution proportion is greater than 0.55 (seeing Table 1), so think malicious attack.To " Lena " in the watermarking images 10 figure, hair will be distorted in the white line zone, and tampering methods is for strengthening brightness, and 19 for distorting effect; Can find also positioning tampering from corresponding watermark extracting Figure 15 and watermark difference Figure 16.A dense as calculated distribution proportion is greater than 0.55 (seeing Table 1), so think malicious attack equally.
In the embodiment of Fig. 6 E-seal, be that the original watermark of 256 * 256 pixels carries out the resolution decrement operations at first to size, obtain watermark pyramid 11, top layer is to following, and resolution increases progressively successively, and layered image length and width direction number of picture elements is followed successively by 32 * 32,64 * 64,128 * 128.Be that the original seal image of 256 * 256 pixels carries out 3 layers of wavelet decomposition to size then, 1,2, the 3 layers of corresponding respectively subband that is embedded into wavelet coefficient 7,8,9 of seal pyramid that will dewater then are as Fig. 2.Embed algorithm as described in the preceding summary of the invention, be described below in conjunction with this embodiment:
(1) get m=4, promptly each wavelet coefficient embeds the watermark information of 4bit.In conjunction with the characteristics that embed algorithm and the size of watermark pyramid resolution, we are extended to 4bit with the watermark information of 1bit, be about to " 1 " and be extended to " 1100 ", " 0 " is extended to " 0000 ", reduce to greatest extent because the watermark flase drop that error caused of wavelet transformation and inverse transformation.
(2) with the coefficient that quantizes on the matrix quantization respective sub-bands, according to 1. 2. reaching 3. formula judgement and revising corresponding wavelet coefficient.
(3) obtain the watermarking images 10 of E-seal through wavelet inverse transformation.
Seal watermarking images 10 can effectively be resisted content tampering and attack.Such as utilizing image processing software such as Photoshop etc., made the literal in the seal " white one-tenth " into " section's one-tenth ", obtain distorting seal 18.The concrete grammar of verifying this seal be according to the step of Fig. 3 extract watermark extracting Figure 15.Utilization is distorted and is detected taxonomic methods 17 again, and obtaining conclusion is malicious attack, judges that seal is false.
In actual applications, E-seal is applied in the electronic government affairs system in conjunction with digital signature technology.Digital signature protection official document content is not distorted, and E-seal then provides the person's of affixing one's seal authentication, guarantees the legitimacy and the authority of official document.In this electron-like government system, original watermark will be stored in the data in server storehouse, by extracting the watermark in the E-seal, with the original watermark comparison in the database, can verify the authenticity of this seal, and then judge the authenticity of this seal place official document.
In the accidental attack detecting embodiment of Fig. 7, from left to right respectively correspondence to stand compressibility factor be 90%, 80%, 70%, 60%, the watermark extracting figure after 50% the JPEG compression.Distort classification results and see Table 1.
Table 1 is to distort the classification result of determination.Each parameter defines as mentioned in the table 1, has tested JPEG compression and zone respectively and has revised two kinds of distort the most common.Result of determination shows that the zone modification is considered to malice and distorts.To can be as small as compressibility factor be 70% and the present invention resists JPEG compression.
The attack classification of table 1 pair above-mentioned attack is judged table
Attack | The JPEG compression | Revise in the zone | |||||
????90% | ????80% | ????70% | ????60% | ????50% | Replace | Blast | |
??λ 1 | ????0.02 | ????0.12 | ????0.30 | ????0.40 | ????0.44 | ????/ | ????/ |
??λ 2 | ????0.32 | ????0.44 | ????0.46 | ????0.49 | ????0.49 | ????/ | ????/ |
??λ 3 | ????0.23 | ????0.48 | ????0.49 | ????0.50 | ????0.50 | ????/ | ????/ |
??δ 1 | ????0.10 | ????0.25 | ????0.42 | ????0.92 | ????0.93 | ????1 | ????1 |
??δ 2 | ????0.87 | ????0.93 | ????0.95 | ????0.96 | ????0.96 | ????1 | ????1 |
??δ 3 | ????0.80 | ????0.98 | ????0.99 | ????0.99 | ????0.99 | ????1 | ????1 |
Attack and judge | The accidental modification | The accidental modification | The accidental modification | Malicious attack | Malicious attack | Malicious attack | Malicious attack |
Table 2 quantization threshold matrix table
The wavelet coefficient zone | ??LL 3 | ??LH 3 | ??HL 3 | ??HH 3 | ??LH 2 | ??HL 2 | ??HH 2 | ??LH 1 | ??HL 1 | ??HH 1 |
Quantization threshold JND (i, j) | ?11.36 | ??12.71 | ??12.71 | ??19.55 | ??14.69 | ??14.69 | ??28.43 | ??23.04 | ??23.04 | ??58.43 |
Claims (4)
1, a kind of fragile watermark method that is used for E-seal protection is characterized in that this method is made up of four steps: 1) construct the original watermark pyramid; 2) in conjunction with human visual system's embed watermark; 3) watermark extracting; 4) multiresolution is distorted and is detected and authentication.
2, fragile watermark method according to claim 1 is characterized in that this method step 1) structure original watermark pyramid is the method for successively decreasing according to the resolution that the JBIG of bianry image expert group Standard of image compression provides, and divides three levels to construct.
3, fragile watermark method according to claim 1, it is characterized in that this method step 2) in conjunction with human visual system's embed watermark, be to adopt the quantization matrix threshold value in wavelet transformed domain, the wavelet coefficient of original image to be done to quantize modulation, so that detect any content tampering that visually can differentiate.
4, fragile watermark method according to claim 1 is characterized in that this method step 4) multiresolution distorts detection and undertaken by distorting classifying rules, according to calculating the difference figure D that extracts between watermark and the original watermark
l(i, j)=| W
l(i, j)-W
l' (i, j) | judge:
(1) for any l, λ
l=0, there is not modification so; (2), make λ if there is l
l>0 and, δ
l<α thinks accidental modification so; (3) for any l, if δ
l〉=α thinks malicious attack; Wherein l is a level of resolution, λ
l, δ
lBe respectively that the detection erroneous point of distorting of watermark embedding subband is densely distributed ratio shared in this subband is had a few with distorting to detect in the erroneous point under the level of resolution l, α is the real number between predefined 0.5 to 1, generally is taken as 0.55.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200310111928 CN1234230C (en) | 2003-10-29 | 2003-10-29 | Vulnerable watermark method for protecting electronic signet |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200310111928 CN1234230C (en) | 2003-10-29 | 2003-10-29 | Vulnerable watermark method for protecting electronic signet |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1540599A true CN1540599A (en) | 2004-10-27 |
CN1234230C CN1234230C (en) | 2005-12-28 |
Family
ID=34336302
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 200310111928 Expired - Fee Related CN1234230C (en) | 2003-10-29 | 2003-10-29 | Vulnerable watermark method for protecting electronic signet |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN1234230C (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100346353C (en) * | 2005-07-19 | 2007-10-31 | 电子科技大学 | Generating and vertification method for electronic seal based on vulnerable water mark |
CN104091302A (en) * | 2014-07-10 | 2014-10-08 | 北京工业大学 | Robust watermark embedding and extracting method based on multi-scale space |
CN105912894A (en) * | 2016-04-07 | 2016-08-31 | 广西国盾科技有限公司 | Method for applying certificate number to electronic seal stamp anti-counterfeiting |
CN110489943A (en) * | 2019-07-15 | 2019-11-22 | 贝壳技术有限公司 | Page watermark tamper resistant method and system |
-
2003
- 2003-10-29 CN CN 200310111928 patent/CN1234230C/en not_active Expired - Fee Related
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100346353C (en) * | 2005-07-19 | 2007-10-31 | 电子科技大学 | Generating and vertification method for electronic seal based on vulnerable water mark |
CN104091302A (en) * | 2014-07-10 | 2014-10-08 | 北京工业大学 | Robust watermark embedding and extracting method based on multi-scale space |
CN104091302B (en) * | 2014-07-10 | 2017-06-06 | 北京工业大学 | A kind of robust watermarking insertion and extracting method based on multiscale space |
CN105912894A (en) * | 2016-04-07 | 2016-08-31 | 广西国盾科技有限公司 | Method for applying certificate number to electronic seal stamp anti-counterfeiting |
CN105912894B (en) * | 2016-04-07 | 2018-11-23 | 广西国盾科技有限公司 | A method of it is anti-fake that passport NO. being used for E-seal printed text |
CN110489943A (en) * | 2019-07-15 | 2019-11-22 | 贝壳技术有限公司 | Page watermark tamper resistant method and system |
Also Published As
Publication number | Publication date |
---|---|
CN1234230C (en) | 2005-12-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101059863A (en) | Embed and detection method for identifying water mark, its system and uses | |
CN101042769A (en) | Active mode digital image content identification method based on wavelet and DCT dual domain | |
CN1885341A (en) | Reversible watermark method for image certification | |
Khan et al. | Robust method for detection of copy-move forgery in digital images | |
CN1256705C (en) | Small wave region digital water marking mathod based on image target region | |
Wang et al. | Wavelet based region duplication forgery detection | |
CN102279969B (en) | Print-scan resistant digital watermark method based on contourlet and quotient space | |
CN103079066B (en) | Image watermark embedding and extracting method combining ridgelet transform and DCT transform | |
CN103927711A (en) | Robust digital watermark algorithm based on OFDM-CDMA | |
CN1246779C (en) | Embedding and extracting method for wavelet domain chaotic semi-frageile digit water print | |
CN1234230C (en) | Vulnerable watermark method for protecting electronic signet | |
CN1251140C (en) | Encrypting orthogonal transformation method for digital watermark | |
Giakoumaki et al. | Multiple digital watermarking applied to medical imaging | |
Li et al. | Blind image watermarking scheme based on wavelet tree quantization robust to geometric attacks | |
Maity et al. | An image watermarking scheme using HVS characteristics and spread transform | |
CN1414778A (en) | Watermark method using geometry calibrated anti geometry conversion image | |
CN1148961C (en) | Method for embedding digital watermark | |
Bashar et al. | Wavelet-Based Multiresolution Features for Detecting Duplications in Images. | |
Gu et al. | Multipurpose image watermarking in DCT domain using subsampling | |
CN1664849A (en) | Image identification method based on parameter integer wavelet transformation | |
Xie et al. | A new fusion based blind logo-watermarking algorithm | |
CN100340111C (en) | Medical image fragile watermark method based on wavelet transform | |
Hosam et al. | A hybrid ROI-embedding based watermarking technique using DWT and DCT transforms | |
CN1889672A (en) | Video-frequency identifying method based on small wave conversion and mixed watermark | |
SEO et al. | A digital watermarking algorithm using correlation of the tree structure of DWT coefficients |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
C17 | Cessation of patent right | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20051228 Termination date: 20091130 |