CN1421782A - Digital watermark anti-fake method for bills - Google Patents

Abstract

The digital watemrmark anti-fake method for bills includes integral checking of bill information, random interrupting, filling, one-dimensional DCT conversion, Rijindael enciphering, region interrupting UUEncode to obtain anti-fake code with hidden digital watermark. The digital watermark anti-fake method has limited length plain text, and thus has greatly raised system speed, reduced system software and hardware overhead. The integral checking can find out all the one-digital altering and continuous two-digit altering to protect integral data; and adopting one-dimensional DCT conversion and Rijindael enciphering of 128 bit cipher key can resist all attack to ensure the safety of data information in enciphere bill.

Description

A kind of digital watermark anti-fake method of bill

Technical field:

The present invention relates to a kind of digital watermark anti-fake method of bill.

Background technology:

Along with popularizing of the arrival of information age, particularly Internet, the safeguard protection problem of information becomes increasingly conspicuous.Current information security technology is basically based on the cryptography theory; its protected mode all is accesses of control documents; be about to file encryption and become ciphertext; the disabled user can not be understood; along with the widespread use of multimedia technology, need encrypt, the acoustic image data of authentication and copyright protection are also more and more.Digitized acoustic image data are exactly digital signal from essence, if this class signal is also adopted the mode of password encryption, then the signal attribute of itself just has been left in the basket.Recent years, many researchists have abandoned traditional cryptographic technology path, attempt adopting various method for processing signals that the acoustic image data are carried out hidden encryption, and this technology is applied to make multimedia " digital watermarking ".

Digital watermark technology is meant with method for processing signals and embeds implicit mark that in digitized multi-medium data this mark is normally sightless, has only detecting device or reader by special use to extract.Design of Digital Watermark must satisfy the disabled user can not be discerned, and is merely able to by the legal requirement that the copyright owner confirmed.

The digital watermark processing technology can be divided into two big classes: time domain watermark processing technology and transform domain watermark processing technology.The advantage of time domain watermark processing technology is that the speed of the embedding of watermark and detection is fast, shortcoming is that attacking abilities such as antinoise, filtering, compression are poor, and embed watermark information is little, in transform domain watermark processing technology, (---DCT or wavelet transform---DWT) as discrete Fourier transform (DFT), discrete cosine transform at first the sampled data of digital signal to be carried out certain conversion.Associative transformation self characteristics and people's requirement is embedded into watermark information on the coefficient in transform domain of digital signal then, recovers the digital signal that contains watermark information by inverse transformation at last.The advantage of transform domain watermark processing technology is containing much information of embed watermark, and attacking abilities such as antinoise, filtering, compression are more intense, and the robustness of embed watermark is good, and transform domain watermark processing technology is the main flow research direction of present digital watermark processing technology.But another aspect, existing transform domain digital watermark method can not be implemented the harmless expression of digital signal, therefore can not be applied to the false proof of bill.

Summary of the invention:

The object of the present invention is to provide a kind of digital watermark anti-fake method of bill, this method can realize the harmless reconstruct of image, and have good security owing to adopted one dimension integer dct transform and the Rijndael encryption.

For achieving the above object, the step of adding watermark signal is in data image signal:

1), binary data stream to be encrypted is done CRC check one time by checker, generate the verification nuclear of two byte longs, binary data stream to be encrypted is intercepted 8n-2 byte data at random, n 〉=2, if it is full that the data intercept byte number is then filled with 0 or 1 inadequately, and high-ranking officers check the back be added on institute's data intercept stream, generate plaintext to be encrypted;

2), with plaintext input frequency domain transducer to be encrypted, the frequency domain transform device will be expressly by 8 segmentations, and make one dimension integer dct transform to every section, obtain the IntDCT coefficient;

3), the watermark inserter joins the intermediate-frequency section of IntDCT coefficient with digital watermark information, obtains containing the IntDCT coefficient of watermark information;

4), the encryption equipment IntDCT coefficient that will the contain watermark information Rijndael that carries out 128 keys encrypts, and generates ciphertext;

5), ciphertext imported randomizer carry out zone intercepting UUEncode, make it can show as discernible alphabetical information, obtained the security code of implicit digital watermarking at par at last.

Extract the step of watermark signal:

1), the security code that will imply digital watermarking is input to demoder and carries out anti-zone intercepting UUEncode, obtains ciphertext;

2), ciphertext is imported decipher carries out anti-Rijndael deciphering, output IntDCT coefficient;

3), the IntDCT coefficient is input to the watermark extracting device, obtain digital watermarking.

The advantage that the present invention had: A, adopt time-limited plaintext, can improve the speed of system greatly, and reduce that system is soft, the spending of hardware; B, the whole verification of employing can be checked out one all changes and two continuous changes, realize the protection to overall data; C, employing one dimension integer dct transform and key length are that 128 Rijndael encrypts, and can resist all attacks, guarantee the safety of ciphered data information.

The present invention is further illustrated below in conjunction with accompanying drawing.

Description of drawings:

Accompanying drawing 1 is the theory diagram that adds watermark information in digital signal.

Accompanying drawing 2 is for extracting the theory diagram of watermark information from the digital signal that contains watermark signal.

Accompanying drawing 3 is first kind 8 integer DCT process flow diagrams.

Accompanying drawing 4 is second kind of integer DCT process flow diagram.

Embodiment:

Referring to Fig. 1; watermark information to be inserted in the binary data stream to be encrypted; at first binary data stream to be encrypted is done CRC check one time by checker 1; to prevent the illegal binary data stream to be encrypted of changing; and binary data stream to be encrypted made Global Macros; generate the verification nuclear of two byte longs; binary data stream to be encrypted is intercepted 14 byte datas at random (also be chosen as 22; 30 bytes etc.; byte is short more; then system speed is fast more; and can reduce the soft of system; the hardware spending); if not enough 14 bytes are then filled 0 (or 1) to 14 bytes; and high-ranking officers check the back be added on 14 byte datas that intercept, generate 16 bytes plaintext to be encrypted.

Above-mentioned CRC check principle is as follows:

In fact the CRC inspection principle is exactly an additional r position scale-of-two check code (sequence) after a p bit binary data sequence, thus the binary sequence that to constitute a length overall be the n=p+r position, for example, p bit binary data sequence D=[d _P-1d _P-2... d ₁d ₀], r position scale-of-two check code R=[r _R-1r _R-2... .r ₁r ₀], resulting this n position binary sequence is exactly M=[d _P-1d _P-2... d ₁d ₀r _R-1r _R-2... .r ₁r ₀]; Be attached between the content of this check code after the data sequence and data sequence and exist certain specific relation.If make a mistake because of reasons such as interference make a certain position or some position in the data sequence, this particular kind of relationship will be destroyed, therefore, by checking this relation, just can realize the check to the data correctness.

Check code R gets the residue computing and obtains by the data sequence D being carried out binary division, and it is by (r+1) position binary sequence G=[g that is called generator polynomial _rg _R-1... .g ₁g ₀] remove, be expressed as with polynomial form $\frac{x^{r}D\left(x\right)}{G\left(x\right)}=Q\left(x\right)+\frac{R\left(x\right)}{G\left(x\right)}----\left(1\right)$

Wherein, x ^rD (x) expression is with the data sequence D r position (promptly increasing r 0 at the end of D again) that moves to left, and Q (x) represents the merchant of this division gained, and R (x) is exactly required residue.This operation relation can also be expressed with formula (2) $R\left(x\right)=\mathrm{Re}\left[\frac{x^{r}D\left(x\right)}{G\left(x\right)}\right]----\left(2\right)$ Wherein, Re[] represent the formula in the bracket is got the residue computing.The coding of check code calculates as mentioned above, and checkout procedure then is the M sequence directly to be carried out division get the residue computing, promptly $\frac{M\left(x\right)}{G\left(x\right)}=Q\left(x\right)+\frac{R\left(x\right)}{G\left(x\right)}----\left(3\right)$

Or be expressed as $R\left(x\right)=\mathrm{Re}\left[\frac{M\left(x\right)}{G\left(x\right)}\right]----\left(4\right)$

Resulting residue R (x) is if zero expression data is correct, otherwise thinks and make a mistake.

The 16 bytes plaintext to be encrypted of checker 1 output is delivered to frequency domain transform device 2, and frequency domain transform device 2 plaintext that 16 bytes are to be encrypted is by 8 segmentations, and makes one dimension integer dct transform to every section, obtains the IntDCT coefficient.

The algorithm of above-mentioned one dimension integer dct transform is as follows:

The transformation matrix C of 8 IntDCT-II of the first kind ₈ ¹¹Be defined as ${{\stackrel{\‾}{C}}_8}^{11}={P_8}^{\left(1\right)}\left[\begin{array}{cc}{I}_4& \\ & J_4\end{array}\right]\left[\begin{array}{ccc}{I}_5& & \\ & {\stackrel{\‾}{R}}_{-\frac{\mathrm{\π}}{4}}& \\ & & 1\end{array}\right]\left[\begin{array}{cc}{P_4}^{\left(1\right)}& \\ & {P_4}^{\left(1\right)}\end{array}\right]\left[\begin{array}{cccccc}{I}_2& & & & & \\ & J_2& & & & \\ & & & 1& 1& \\ & & & 1& -1& \\ & & & & -1& 1\\ & & & & 1& 1\end{array}\right]$ $x\left[\begin{array}{ccc}{\stackrel{\‾}{R}}_{\frac{\mathrm{\π}}{4}}& & \\ & {\stackrel{\‾}{R}}_{\frac{\mathrm{\π}}{8}}& \\ & & {\stackrel{\stackrel{-}{~}}{T}}_4\end{array}\right]\left[\begin{array}{cccccc}1& & & 1& & \\ & -1& -1& & & \\ 1& & & -1& & \\ & -1& 1& & & \\ & & & & I_2& \\ & & & & & I_2\end{array}\right]\left[\begin{array}{cc}{I}_4& J_4\\ I_4& -J_4\end{array}\right],\left(8\right)$

And the transformation matrix C of the second class IntDCT-II ₈ ¹¹Can be defined as again ${C_8}^{11}={P_8}^{\left(1\right)}\left[\begin{array}{ccc}{I}_4& & \\ & & -J_2\\ & J_2& \end{array}\right]\left[\begin{array}{cc}{P_4}^{\left(1\right)}& \\ & {\stackrel{\stackrel{-}{~}}{T}}_4\end{array}\right]\left[\begin{array}{cccccc}{I}_2& & & & & \\ & J_2& & & & \\ & & 1& 1& & \\ & & 1& -1& & \\ & & & & I_2& \\ & & & & & {-I}_2\end{array}\right]\left[\begin{array}{ccc}{\stackrel{-}{R}}_{\frac{\mathrm{\π}}{4}}& & \\ & {\stackrel{-}{R}}_{\frac{\mathrm{\π}}{8}}& \\ & & I_4\end{array}\right]$ $x\begin{array}{c}\left[\begin{array}{ccccc}1& & & 1& \\ & -1& -1& & \\ 1& & & -1& \\ & -1& 1& & \\ & & & & {P_4}^{\left(1\right)}\end{array}\right]\left[\begin{array}{ccc}{I}_5& & \\ & {\stackrel{\‾}{R}}_{\frac{\mathrm{\π}}{4}}& \\ & & 1\end{array}\right]\left[\begin{array}{cc}{I}_4& I_4\\ I_4& {-J}_4\end{array}\right],\left(9\right)\end{array}$

Utilize non-linear nature in the definition of integer transform, for the arbitrary integer sequence, the output result of above-mentioned two kinds of IntDCT-II is generally different, therefore can choose suitable IntDCT-II as required in practical problems.Accompanying drawing 3 and accompanying drawing 4 have provided the process flow diagram of above-mentioned two kinds of 8 IntDCT-II respectively.Wherein, ${{\tilde{I}}_2}^{\left(1\right)}=\mathrm{diag}\{1,-1\},{{\tilde{I}}_2}^{\left(2\right)}=\mathrm{diag}\{-\mathrm{1,1}\},{{\tilde{I}}_2}^{\left(3\right)}=\mathrm{diag}\{-1,-1\}.$

Be generalized to general N=2 with 8 IntDCT-II and IntDCT-IV now with above-mentioned 4 ^t, the design of t 〉=3 IntDCT-II and IntDCT-IV.

Notice matrix I in the proposition 1 _M, V _M, P _M, P _M ⁽¹⁾, D, D _M, J _MThis is as the conversion of integer to integer.

The rank matrix

Can be used for recursive definition IntDCT, and matrix Comprised The rotation of individual plane, matrix

Can obtain by integer transform is approximate.For definition and the algorithm of the IntDCT that sets up general length, remaining is need be to matrix B now _NIn a sub-block matrix

It is approximate to make integer transform, the lemma below for this reason at first setting up.

Lemma 1 matrix

There are following two kinds of decomposed forms $\mathrm{diag}(\sqrt{2},\sqrt{2})=R_{\frac{\mathrm{\π}}{4}}\left[\begin{array}{cc}1& 1\\ -1& 1\end{array}\right],\left(10\right)$

Perhaps $\mathrm{diag}(\sqrt{2},\sqrt{2})=\left[\begin{array}{cc}2& \sqrt{2}-2\\ 0& 1\end{array}\right]\left[\begin{array}{cc}1& \\ 1& 1\end{array}\right]\left[\begin{array}{cc}1& \sqrt{2}-1\\ & 1\end{array}\right]\left[\begin{array}{cc}1& 0\\ -\frac{1}{\sqrt{2}}& 1\end{array}\right],\left(11\right)$

(10) can directly set up with (11), as space is limited, omit herein by matrix multiplication.

Because the matrix that occurs in (10) or (11) is INTEGER MATRICES or has the matrix that promotes structure, so matrix

Can be similar to by integer transform, establishing its integer transform is K ₂

Length N=2 are set up to recursion in comprehensive above-mentioned discussion ^t, the IntDCT-II of t 〉=3 and IntDCT-IV matrix are as follows: ${\stackrel{\‾}{C}}_N^{\mathrm{II}}=P_N^{\left(1\right)}\left[\begin{array}{cc}{I}_{N/2}& 0\\ 0& J_{N/2}\end{array}\right]\left[\begin{array}{cc}{\stackrel{\‾}{C}}_{N/2}^{\mathrm{II}}& \\ & {\stackrel{\‾}{C}}_{N/2}^{\mathrm{IV}}\end{array}\right]\left[\begin{array}{cc}{I}_{N/2}& J_{N/2}\\ I_{N/2}& -J_{N/2}\end{array}\right].\left(4.12\right)$ ${\stackrel{\‾}{C}}_N^{\mathrm{IV}}=P_N^{\left(2\right)}{\stackrel{\‾}{B}}_N{P}_N^{\left(2\right)}{P}_N^{\left(1\right)}\left[\begin{array}{cc}{\stackrel{\‾}{C}}_{N/2}^{\mathrm{II}}& \\ & {\stackrel{\‾}{C}}_{N/2}^{\mathrm{II}}\end{array}\right]\left[\begin{array}{cc}{I}_{N/2}& \\ & D_{N/2}{J}_{N/2}\end{array}\right]{\stackrel{\‾}{\tilde{T}}}_N\left[\begin{array}{cc}{I}_{\frac{N}{2}}& \\ & -I_{\frac{N}{2}}\end{array}\right],\left(13\right)$

Wherein, B _N=diag (K ₂, V _N-2), Comprised The integer lifting matrix decomposition of individual rotational transform.Certainly, utilize the symmetry of DCT-IV, can also be as 8 IntDCT-IV, define multiple integer DCT-IV and corresponding multiple integer DCT-II. reaches following algorithm with above-mentioned discussion list now.

Algorithm (1). length N=2 ^t, the calculating of the IntDCT-II of t 〉=3: establishing list entries is x (k), k=0,1,2, A, N-1.

The first step: calculate $g\left(k\right)=x\left(k\right)+x(N-k-1),h\left(k\right)=x\left(k\right)-x(N-k-1),k=\mathrm{0,1,2},\mathrm{\Λ},\frac{N}{2}-1;$

Second step: difference sequence of calculation g (k) and sequences h (k)

Point IntDCT-II and IntDCT-IV establish it and are output as $H_1\left(k\right),0\≤k\≤\frac{N}{2}-1$ With $H_1\left(k\right),\frac{N}{2}\≤k\≤N-1.$

The 3rd step: sequence is reset $X=P_N^{\left(1\right)}\left[\begin{array}{cc}{I}_{N/2}& 0\\ 0& J_{N/2}\end{array}\right][H_1,\left(0\right),H_1\left(1\right),\mathrm{\Λ},H_1(N-1){]}^{\′}.$

Above-mentioned three steps carry out according to length N recursion ground.

(2). length N=2 ^t, the calculating of the IntDCT-IV of t 〉=3: establishing list entries is x (k), k=0,1,2, Λ, N-1.

The first step: for sequence $x\left(k\right),0\≤k\≤\frac{N}{2}-1$ And $-x\left(k\right),\frac{N}{2}\≤k\≤N-1,$ According to $H=[h\left(0\right),h\left(1\right),\mathrm{\Λ},h(N-1){]}^{\′}={\stackrel{\‾}{\tilde{T}}}_N[x\left(0\right),x\left(1\right),\mathrm{\Λ},x(\frac{N}{2}-1),-x\left(\frac{N}{2}\right),$ $\mathrm{\Λ},-x(N-1)]{\′}^{}$ Do to promote conversion;

Second step: reset for the latter half of of sequence H: $[h\left(\frac{N}{2}\right),h(\frac{N}{2}+1),\mathrm{\Λ},h(N-1){]}^{\′}:=[h(N-1),-h(N-2),h(N-3),-h(N-4),\mathrm{\Λ},-h\left(\frac{N}{2}\right){]}^{\′}$

Calculate respectively then $h\left(k\right),0\≤k\≤\frac{N}{2}-1$ With $h\left(k\right),\frac{N}{2}\≤k\≤N-1$

Point IntDCT-II establishes it and is output as $H_2\left(k\right),0\≤k\≤\frac{N}{2}-1$ With $H_2\left(k\right),\frac{N}{2}\≤k\≤N-1;$

The 3rd step: according to $Y=P_N^{\left(2\right)}{P}_N^{\left(1\right)}[H_2\left(0\right),H_2\left(1\right),\mathrm{\Λ},H_2(N-1){]}^{\′}$ Do sequence and reset, calculate then, $\left[\begin{array}{c}y\left(2\right)\\ y(N-1)\end{array}\right]:=\left[\begin{array}{cc}1& 1\\ -1& 1\end{array}\right],\left[\begin{array}{c}y(2k+1)\\ y(2k+2)\end{array}\right]:=\left[\begin{array}{cc}1& 1\\ -1& 1\end{array}\right]\left[\begin{array}{c}y(2k+1)\\ y(2k+2)\end{array}\right],k=\mathrm{1,2},N/2-2;$

The 4th step: according to $X=P_N^{\left(2\right)}Y$ Do the sequence rearrangement and obtain needed IntDCT-IV sequence.

Notice that some matrix itself is exactly an INTEGER MATRICES in the definition of IntDCT, and it for example has against being INTEGER MATRICES also except that constant factor: $V_M^{-1}=\frac{1}{2}{V}_m,P_N^{{\left(1\right)}^{-1}}=P_M^{{\left(1\right)}^{\′}},P_N^{{\left(2\right)}^{-1}}=P_N^{\left(2\right)},J_N^{-1}=J_N,{\left(D_M{J}_M\right)}^{-1}=D_M{J}_M$ And $\left[\begin{array}{cc}{I}_{\frac{N}{2}}& J_{\frac{N}{2}}\\ I_{\frac{N}{2}}& {-J}_{\frac{N}{2}}\end{array}\right]=\frac{1}{2}\left[\begin{array}{cc}{I}_{\frac{N}{2}}& I_{\frac{N}{2}}\\ J_{\frac{N}{2}}& {-J}_{\frac{N}{2}}\end{array}\right].$

After frequency domain transform device 2 carried out one dimension integer dct transform, watermark inserter 3 joined the intermediate-frequency section of IntDCT coefficient with watermark information, obtained containing the IntDCT coefficient of watermark information like this; It is that 128 Rijndael encrypt that the IntDCT coefficient that encryption equipment 4 will contain watermark information carries out key length, and generates ciphertext; The ciphertext that generates is sent to UUEncode randomizer 5, adopts the UUEncode technology to carry out zone and intercepts, and it is that binary data stream is all split into the position is the data stream of unit, is a as adjacent two bytes, b.Then the result of Chai Fening is a ₀a ₁a ₂a ₃a ₄a ₅a ₆a ₇b ₀b ₁b ₂b ₃b ₄b ₅b ₆b ₇

A wherein _i, b _i=0,1} (i=0,1 ... .7).It is that a sign indicating number is formed in a unit that the contraposition data stream is chosen 6 successively, its span is 0-63, with it as sequence number, in given character set the inside value, make it can show as discernible alphabetical information, obtained the security code of implicit digital watermarking at par at last.

Above-mentioned encryption equipment 4 also can adopt other known cryptographic algorithm.

Referring to Fig. 2, the process of extracting watermark signal from the security code of implicit digital watermarking is: the security code that will imply digital watermarking is input to demoder 6 and carries out anti-zone intercepting UUEncode, obtains ciphertext; Ciphertext is delivered to decipher 7 carry out anti-Rijndael deciphering, export the IntDCT coefficient of moisture official seal breath; The IntDCT coefficient of moisture official seal breath is delivered to watermark extracting device 8, obtain digital watermarking.

Claims (4)

1. method that adds watermark signal in bill numerical information is characterized in that: may further comprise the steps into:

A, the binary data stream of bill to be encrypted is done CRC check one time by checker (1), generate the verification nuclear of two byte longs, binary data stream to be encrypted is intercepted 8n-2 byte data at random, n 〉=2, if it is full that byte is then filled with 0 inadequately, and high-ranking officers check the back be added on data intercept stream, generate plaintext to be encrypted;

B, with plaintext input frequency domain transducer (2) to be encrypted, frequency domain transform device (2) will be expressly by the segmentation of (8) point, and makes one dimension integer dct transform to every section, obtains the IntDCT coefficient;

C, watermark inserter (3) join the intermediate-frequency section of IntDCT coefficient with digital watermark information, obtain containing the IntDCT coefficient of watermark information;

D, encryption equipment (4) will contain the IntDCT coefficient encryption of watermark information, generate ciphertext;

E, ciphertext is imported randomizer (5), adopt the UUEncode technology to carry out zone and intercept, make it can show as discernible alphabetical information, obtained the security code of implicit digital watermarking at par at last.

2. a kind of method that adds watermark signal in bill numerical information according to claim 1 is characterized in that: 14 byte datas of intercepting among the described step a.

3. a kind of method that adds watermark signal in bill numerical information according to claim 1 and 2, it is characterized in that: described encryption equipment (4) is the Rijndael encryption equipment.

4. method of extracting watermark information from the security code of implicit digital watermarking is characterized in that: may further comprise the steps:

A, the security code that will imply digital watermarking are input to demoder (6) and carry out the intercepting of anti-zone, obtain ciphertext;

B, ciphertext is imported decipher (7) carry out anti-Rijndael deciphering, output IntDCT coefficient;

C, with IntDCT coefficient input watermark extracting device (8), obtain digital watermarking.

Images