CN1527585A - Safe digital information transmitting and applying method - Google Patents

Abstract

The present invention is safe digital information transmitting and applying method based on digital water mark technology and belongs to the field of mode identification and information safety. By means of spatial domain water mark algorithm based on orthogonal template, water mark information is added into image, video or audio data, and the algorithm is simple, reliable and robust. By means of spatial domain water mark algorithm based on two valued and multiple valued discrete image, water mark information may be embedded into table, text, stamp, hand signature, curve and other discrete image. By means of the two said algorithms and biological characteristic information, the biological characteristic information may be embedded into certificate, bill, ID card and other media in water mark form.

Description

A kind of digital information safety transmission and methods for using them

The present invention is a kind of message transmission and application process based on digital watermark technology, belongs to pattern recognition and information security field.

The appearance of network and computer, scanner and printer widely-used makes obtaining, exchanging of digital information and transmission becomes simple unusually." but digital world " is still stranger for ordinary people's daily life.General people still more be ready by traditional non-electronic medias such as printed article carry out that information interchange, certificate transmit, contract signing or the like, though the message transmission of digital world safe enough.How the convenience of traditional media being combined with fail safe on the electronic media, is the difficult problem that needs solve.Digital watermark technology then provides an effective solution for this difficult problem.

Digital watermarking is to add some digital information to reach effect such as copyright protection in multi-medium data (as image, sound, vision signal etc.).General digital watermarking algorithm comprises watermark embedding and watermark extracting or detects this two basic sides.Kind from algorithm, watermarking algorithm can be divided into two big classes basically: class methods are by directly the be added to spatial domain (Spatial Domain) of image of certain algorithm with digital watermarking, another kind of method is earlier image to be done certain conversion (particularly orthogonal transform), then watermark is embedded into the transform domain (TransformDomain) of image.From present case, it is more prevalent that the transform domain method is just becoming.Because the transform domain method all has good robustness usually, image compression, image filtering and noise pollution commonly used all there is certain resistance.But the advantage of spatial domain method is that its algorithm is simple, and computational speed is than very fast.Though digital watermark technology is at digital media, a good watermarking algorithm is equally applicable to nonnumeric media, promptly passes through " digital-to-analog/numeral " such conversion, and watermark information still can intactly be extracted or detect.Digital watermarking can make the information security technology that we enjoy much can only just have at present as an information carrier in digital world on traditional media so.

The technical literature that can contrast has following five pieces:

[1]I.J.Cox，J.Kilian，T.Leighton，and?T.Shamoon，SecureSpread?Spectrum?Watermarking?for?Multimedia，NEC?ResearchInstitute?Technical?Report?95-10，1995.

[2]A.Piva，M.Barni，F.Bartolini?and?V.Cappellini，DCT-basedwatermark?recovering?without?resorting?to?the?uncorrupted?originalImage，Proc.of?ICIP′97，Vol.3，pp.520-523，1997.

[3]J.J.K.O′Ruanaidh，W.J.Dowling?and?F.M.Boland，Watermarking?digital?images?for?copyright?protection，IEEProc.-Vis.Image?Signal?Process，Vol.143，No.4，pp.250-256，August，1996.

[4]F.A.P.Petitcolas，R.J.Anderson?and?M.G..Kuhn，Informationhiding-a?survey，Proc.of?the?IEEE，Vol.87，No.7，pp.1062-1078，1999.

[5]R.Z.Liu?and?T.N.Tan，Watermarking?for?digital?images，Proceedings?of?ICSP′98，Vol.2，pp.944-947，1998.

The present invention proposes a kind of information security transmission and application process based on digital watermarking.In conjunction with asymmetric digital encryption algorithm, as carrier, the present invention can go up transmitting digital information at various media (electronics or non-electronic media) safely and effectively with watermark.The digital information of process encryption or signature as required, can be converted into various by the watermark module as image, figure, form, literal, bar code or the like the form of expression, thereby expanded the application of digital watermarking on non-electronic media greatly, made the watermark various fields such as true and false discriminating of application and certificate false proof, document effectively.

Technical essential of the present invention is:

1: based on the spatial domain digital watermarking algorithm of orthogonal template

Digital watermarking algorithm must be protected two essential parts: the embedded part of watermark and the test section of watermark.Its basic principle as shown in Figure 1.Based on the watermarking algorithm difference of transform domain, the spatial domain algorithm is simple, computational speed is fast.For the coherent detection of watermark, if pre-designed suitable template so also just can be designed the watermarking algorithm of practical robust.

If an images is represented with A, be defined as A ∈ R ^{N * M}, wherein R represents real number field.If T ∈ is R ^{K * K}Be the watermark template, K represents template size (as K=8,16,32 or the like).Consider two factors when design template: the frequency domain components of (1) template should belong to Mid Frequency, promptly can not all be that low frequency component (causes the watermarking images distortion too big, and the coherent detection effect is bad), can not all be high fdrequency component (watermark information is lost easily, particularly passes through the lossy compression method of image); (2) orthogonalization of template design is if promptly there is p orthogonal template (p is the positive integer greater than 1): T ₁, T ₂..., T _p, they should satisfy following condition so

T ₁T ₂...T ₃T _p (1)

Provide an additivity watermarking algorithm below based on template.At first establishing watermark is vectorial W={w ₁, w ₂..., w _i..., w _l, wherein vectorial number $l\≤N/K\×M/K,$ Vector element satisfies w _i∈ 0,1,2 .., p}.For the sake of simplicity, be example with p=1, so watermark vector then be two-value 0, the 1} sequence.Image A is resolved into size be the image block A of K * K _K, auto-correlation energy (being the F norm) and template and the A of difference calculation template T _KCorrelation energy:

$\mathrm{Ep}=\mathrm{\&Sigma;\&Sigma;}{T_{\mathrm{ki},\mathrm{<figure-callout\; id="2"\; label="kj"\; filenames="A0311566100211.png"\; state="\{\{state\}\}">kj</figure-callout>}}}^2,\mathrm{Eb}=\mathrm{\&Sigma;\&Sigma;}{T}_{\mathrm{ki},\mathrm{kj}}\&times;A_{\mathrm{ki},\mathrm{kj}}^K---\left(2\right)$

So to each A _KCarry out following operation:

A _K＝A _K+sT (3)

As each A _KAfter all having calculated, just become the new image that comprises watermark after the combination.The calculating of scale factor s is carried out according to following step:

if?w _i＝1，

if?Eb＞Ep，

s＝0

elseif?Eb＞0?AND?Eb＜Ep

$s=\frac{K\&CenterDot;\mathrm{bm}}{{\left|\right|T\left|\right|}_F}{10}^{-\frac{\mathrm{Dpsnr}}{20}}(1-\mathrm{Eb}/\mathrm{Ep})$

else

$s=\frac{K\&CenterDot;\mathrm{bm}}{{\left|\right|T\left|\right|}_F}{10}^{-\frac{\mathrm{Dpsnr}}{20}}$

endif

endif

if?w _i＝0，

if?Eb＜-Ep，

s＝0

elseif?Eb＜0?AND?Eb＞-Ep

$s=-\frac{K\&CenterDot;\mathrm{bm}}{{\left|\right|T\left|\right|}_F}{10}^{-\frac{\mathrm{Dpsnr}}{20}}(1+\mathrm{Eb}/\mathrm{Ep})$

else

$s=-\frac{K\&CenterDot;\mathrm{bm}}{{\left|\right|T\left|\right|}_F}{10}^{-\frac{\mathrm{Dpsnr}}{20}}$

endif

endif (4)

Wherein bm is the maximum pixel value of image, and Dpsnr allows the distortion factor, ‖ ‖ for the image of making index with PSNR _FExpression F norm.Obviously Dpsnr has determined the size of scale factor s.

The extraction of watermark is finished with coherent detection.To an image B ∈ R to be detected ^{N * M}, be broken down into size and be the image block B of K * K _K, do coherent detection with template then:

${\stackrel{\&OverBar;}{E}}_b=\mathrm{\&Sigma;\&Sigma;}{T}_{\mathrm{ki},\mathrm{kj}}\&times;B_{\mathrm{ki},\mathrm{kj}}^K---\left(5\right)$

If E _b＞0, then watermark information is 1, otherwise, then be 0.

If a plurality of templates are arranged, image block A so _KThen can embed the information of a plurality of bits.Such as being provided with two template T ₁, T ₂, two modules can embed 4 kinds of states so: 0,1,2,3, and template T wherein ₁ Expression 0,1 state, template T ₂Expression 3,4 states.Four kinds of states can be represented with two binary numbers.The rest may be inferred, if 4 templates are arranged, can represent 8 kinds of states so, so available three binary number representations.The prerequisite that this way is set up is that all templates will satisfy formula (1), and is mutually orthogonal.

Usually can produce a small amount of error when detecting watermark, these errors can be resolved by watermark information being carried out the ECC coding fully.The advantage of this algorithm is: do not need original image when the first, detecting watermark, can realize blind Detecting; The second, algorithm compares robust, can resist common printing, scanning and JPEG, MPEG lossy compression method; Three, the watermark capacity of Qian Ruing is big, and the method for designing of multi-template can increase the embedded quantity of watermark information greatly; Four, adaptive scale factor calculation method makes that telescopiny is dynamic, adaptive, satisfies the invisible needs of watermark.

2: at the digital watermarking algorithm of video and audio file

According to top orthogonal template watermarking model, can easily the image watermark algorithm be generalized on video and the audio file.For video file, at first video is decomposed into the image sequence of single frames, design two dimension pattern plate T ∈ R then ^{K * K}, the application image watermarking algorithm is embedded into watermark information in each two field picture, watermark extracting also with the image watermark class of algorithms seemingly, directly extract the watermark information of every two field picture, thereby realize watermark embedding and leaching process video sequence.Mpeg video stream for compressed format then is embedded into watermark in the key frame.

The Audio Watermarking Algorithm structure is also similar with image watermark, but slightly different.Because tonic train is an one-dimensional signal, so orthogonal template also is designed to one dimension, i.e. T ∈ R ^KTemplate length depends on the sample rate of voice data.Telescopiny remains the stack watermarking model, and testing process also is coherent detection, but all calculating and operation all on time domain, finish, therefore it is fast to have speed equally, the algorithm characteristic of simple.

3: based on the discrete digital watermarking model of two-value/multivalue image

In some occasion, to a large amount of two-values or many-valued discrete picture (as literal, form, the black and white image after the scanning, seal, bar code or the like), the invisibility of watermark is less demanding, it is then very high that but watermark capacity and algorithm robustness require, and some is different for the water mark method of the structure of watermarking algorithm and gray level image so.Under the situation that guarantees picture quality,, then need on image area, directly do conversion to the picture element of image for guaranteeing higher robust property.Its basic skills is described below:

Step 1: at first discrete picture is carried out piece by structure and content under the situation of not considering background and divide, block-shaped can be square, round dot or further feature figure, how much picture element decides on concrete application in each piece, picture element is few more, the quantity of piece is just many more, and embeddable information is also many more;

Step 2: each piece can embed the information (image gray levels of piece higher bit number more is also big more, if the piece image has only a gray scale, each piece then can only embed the information of a bit so) of one or more bits according to different needs.Because background is normally white, and human eye is very responsive to image low frequency part (as white background), thereby the background parts (being the higher part of light reflectance, as white " sky ") of image is kept and do not do change.If only embedding a bit with each piece is example, the method for embedding is, if embed watermark information is " 1 ", the pixel value of corresponding blocks is modified and is predefine color value (image as to 256 gray scales is made as 128) so; If embedding information is " 0 ", then this piece keeps original color value, does not do change.

Step 3: according to the visible level of watermark information, (the predefine value is more near the pattern pixel value for the color value of each piece change of predefine, watermark is difficult more to be perceiveed, correspondingly watermark extracting is also difficult more), what (they being the piece image gray levels) of predefine color value then determine watermarked amount of information (a plurality of predefine color values can embed a plurality of bit informations).

4: based on the anti-counterfeit method of credentials of watermark and biometric technology

This method is exactly the anti-counterfeit method of credentials that utilizes digital watermarking, contemporary cryptology and biometric technology to realize.Its core content is:

Step 1: at first biological characteristics such as head portrait are carried out feature extraction, obtain corresponding biological characteristic vector F={f ₁, f ₂..., f _n;

Step 2: to each element f of biological characteristic vector _i, i=1,2 ..., n carries out normalization to be handled.With f _iCodomain be divided into L _iIndividual interval [a _l ⁱ, b _l ⁱ], l=1,2 ..., L _i, define each interval normalization value s _l ⁱ, l=1,2 ..., L _i, make and work as $a_l^l\&le;f_i\&le;b_l^l$ The time $f_l=s_l^i;$

Step 3: the vector of the biological characteristic after the normalization other key message (as name, passport NO., date of issue etc.) with certificate is grouped together, calculating its digital digest then (is the HASH value, as MD5, SHA-1 etc.), and with the key element value of this HASH value as certificate;

Step 4: with the key element value of certificate as raw information, carry out the digital signature processing or make a summary to handle being input to the watermark merge module as watermark, show with the additional clause bar code form then, and be printed in the certificate, thereby finish the antiforge function of certificate.

The present invention and existing digital watermark relatively have following advantage:

At first the present invention proposes multiple bit digital watermark technology based on spatial domain (time domain) orthogonal template.Compare with other digital watermark method, its advantage has the following aspects: (1) all operations is all finished in spatial domain or time domain, orthogonal template is the off-line structure, embeds and detects without any influence for online watermark, so the digital watermarking operation is very fast; Can be done in real time the embedding and the detection of watermark for video and audio file; (2) the watermark amount of Qian Ruing is big, and what of orthogonal template the watermark information that can embed of each image block or audio frequency sheet depend on, embeds the how much irrelevant with the robust performance of watermark of amount of information, does not influence the detection effect of watermark; (3) the unusual robust of algorithm, for the JPEG/MPEG compression, in conjunction with the ECC coding, watermark information can be preserved well.A large amount of experiments show that this algorithm is a very practical robustness watermarking algorithm.

Secondly what the present invention proposed has expanded the application of digital watermarking greatly at the multiple bit digital watermark algorithm of two-value/many-valued discrete picture, this algorithm speed is fast, the containing much information of embedding, and robustness is good, particularly merge mutually, can effectively finish the function of digital anti-counterfeiting with digital signature technology.

The anti-counterfeit method of credentials that the 3rd the present invention proposes based on watermark and biometric technology.Compare with present anti-counterfeit method of credentials, a distinguishing feature of this method is: what (1) certificate comprised is the HASH value of biological information, make that like this certificate can offline inspection, need not the support in network and biological attribute data storehouse, power system capacity is little, good confidentiality is protected user's individual privacy (because detect the HASH value that the certificate true and false only needs biological characteristic, rather than original biological characteristic) effectively; (2) biological information that certificate comprised is to be present in certificate background, digital watermarking bar code, head portrait etc. above the carrier with the watermark form, and document surface is simple for structure clean, and good confidentiality.

Description of drawings:

Fig. 1 is the spatial domain multi-bit watermark algorithm block diagram based on orthogonal template

(a) the incorporation model block diagram of digital watermarking

(b) the detection model block diagram of digital watermarking

Fig. 2 is the example of orthogonal template image watermark algorithm

(a) original image

(b) comprise the image of watermark

Fig. 3 is the block diagram of MPEG video watermark system

(a) digital watermarking of non-compression video sequence embeds block diagram

(b) digital watermarking of compression MPEG video sequence embeds block diagram

(c) digital watermarking of MPEG video sequence detects block diagram

Fig. 4 is the example of MPEG video watermark system

(a) be original video sequence the 102nd frame

(b) be original video sequence the 545th frame

(c) for comprising the 102nd frame of watermark information, do not show watermarking detecting results

(d) for comprising the 545th frame of watermark information, do not show watermarking detecting results

(e) for comprising the 102nd frame of watermark information, show watermarking detecting results

(f) for comprising the 545th frame of watermark information, show watermarking detecting results

Fig. 5 is that the digit water-mark certificate anti-fake that comprises biological characteristic is realized block diagram

Fig. 6 is the exemplary plot of discrete picture watermarking algorithm

(a) original two-value black and white image

(b) the predefine value is 50 watermarking images

(c) the predefine value is 100 watermarking images

(d) the predefine value is 150 watermarking images

(e) the predefine value is 200 watermarking images

(f) adopt the watermarking images of 50,100,150 and 200 4 piece image gray levels simultaneously

Embodiment:

One. based on the spatial domain image watermark system of orthogonal template

How statement now uses the orthogonal template watermarking algorithm to finish the embedding and the extraction of watermark.Here the gray scale image Lena that has provided with 256 * 256 does digital watermarking embedding and the testing result that experiment obtains.The permission distortion factor PSNR value of setting watermarking images is 35dB, and block size is 8, uses an orthogonal template T, and T can be designed to:

1 -1 1 1 -1 -1 1 -1

1 -1 1 1 -1 -1 1 -1

1 -1 1 1 -1 -1 1 -1

1 -1 1 1 -1 -1 1 -1

1 -1 1 1 -1 -1 1 -1

1 -1 1 1 -1 -1 1 -1

1 -1 1 1 -1 -1 1 -1

1 -1 1 1 -1 -1 1 -1

Therefore embeddable maximum watermark capacity is 1024 bits.For during simple, set watermark and be one group at random, have 1024 elements 0, the 1} vector.According to the method that " technical essential one " described, say that the watermark information of 1024 bits is embedded in the image.Fig. 2 provides the result, and wherein Fig. 2-(a) is an original image, and Fig. 2-(b) is the image that contains watermark.

In general, initial p SNR value is set greatly more, and the invisibility of watermark is good more, and it is similar more to original image to contain watermarking images, but the robustness of watermark is also poor more.Original block is selected greatly more, and embeddable watermark capacity is just more little, and the robustness of watermark is good more.

Two .MPEG video watermark systems

Video sequence is to be made of with the still image that waits the space-time distance continuously a series of.Thereby the basic principle of image watermark is applied to video sequence is that of image watermark algorithm promotes naturally.Consider the particularity of video watermark, algorithm only needs to embed very little identification information in implementation procedure, and wishes can resist certain video compression aspect robustness.And just in time can satisfy this two requirements based on the spatial domain image watermark algorithm of orthogonal template.Therefore in the process that embeds, we are embedded into watermark in the key frame; In testing process, use related check and extract watermark content.

Native system is a video watermark implementation at MPEG-1/2.By the video sequence (as digital video camera and image pick-up card etc.) that video capture device obtains, we can carry out the embedding operation of watermark to each frame, and then it is compressed into the MPEG-1/2 video flowing.The manufacturing process of its video flowing is by Fig. 3-(a) provide so.

If given mpeg data stream carries out the processing of digital watermarking if desired to it, our embedding algorithm is mainly at key frame so.Because in MPEG-1/2, the Code And Decode of I frame is independently.Can be extended to sightless video watermark algorithm easily, promptly directly the I frame among the GOP be made the embedding operation of watermark.To each microblocks of I frame, we embed the information of a bit, are 352 * 240 mpeg stream like this to a normal size, and we can embed the amount of information of 1320 bits.This is enough to general application.Fig. 3-(a) is the making block diagram that contains the non-compression video sequence of watermark.Fig. 3-(b) is the block diagram that ready-made MPEG video sequence is carried out watermark processing.

In the testing process of watermark, the video flowing that we will need to verify is input in the decoding video stream device of band watermark detection function, and the output testing result.Its core algorithm also is based on the related detecting method of orthogonal template.Fig. 3-(c) has provided the testing process of video watermark.

The MPEG sequence is the video sequence of one section MPEG-1 form extracting from MPEG-7 standard testing storehouse.In the telescopiny of watermark, we at first decode video, the watermarked sign of application image watermarking algorithm " nlpr " in key frame then, and then it is recovered to the video flowing that contains watermark with mpeg encoder.The mpeg encoder major parameter is set to:

Gop structure is: IBBPBBPBBPBBPBBP, and size is 16 frames;

Frame sign is constant, is 288 * 352;

Bit rate is 1.0MB/s;

YUV sampling is than being 4: 2: 0;

Frame rate is 30f/s;

Fig. 4-(a-b) has provided two two field pictures wherein; (c-d) be corresponding two two field pictures after the watermarked expression " nlpr ".Original video sequence is colored, is raising speed, is translated into gray level image in the coding manufacturing process of mpeg stream.More original and comprise the MPEG-1 video sequence of watermark, both do not have too big vision difference basically.Fig. 4-(e-f) then demonstrates the testing result of watermark.The video sequence that will comprise watermark is input in the mpeg decoder of band watermark detection function, if detected watermark, so watermark content " nlpr " is extracted, and is shown that as sign the percentage in the image lower left corner represents to detect confidence level.Owing in the watermark encoder process, adopted error control coding, so here when detecting confidence level and surpass 50%, we just think the watermark existence.

Three. based on the digital anti-counterfeiting certificate system of face characteristic

This method is utilized technical protections such as digital watermark technology, digital signature, digital digest, living things feature recognition exactly and is prevented that certificate is forged, as shown in Figure 5.

At first 1. certificate information is extracted summary, then summary info is encrypted the back and be embedded into original face picture 2. in (or other position of card, background patterns etc.) as watermark (other certificate identifying information such as certificate sequence number, identification number etc. also can be used as watermark).Extract the biological characteristic vector of face picture, and calculate the HASH value, then and the information 1. utilize the manufacturing terminal private key 4. to do digital signature together;

Secondly, utilize passport NO. to do one-dimensional bar code, then with digital signature as watermark be embedded in the bar code (the optional scheme of another one be directly with the biological characteristic vector of face picture/fingerprint/iris as watermark, rather than do digital signature and be embedded in the bar code.Perhaps on bar code and certificate background, embed digital signature and biological characteristic vector simultaneously);

Print certificate at last 7..

Each part certificate all is a multiple anti-fake like this: (1) all head portraits all are the head portraits that contains watermark, and any certificate that does not comprise the relevant watermark image is all forged; (2) bar code on all indentations one can carry out various management by common bar code typing instrument typing passport NO., judge by the digital signature of extracting in the bar code whether certificate is that original issuing unit signs and issues in addition, all are not all forged by the certificate of digital signature check; (3) digital signature has comprised biological information, and all can not judge that by the certificate of biological information check certificate head portrait and certificate content do not meet.

Four. comprise the bar code of digital watermarking

Fig. 6 provides the experimental result that discrete watermarking algorithm is applied to bar code.The original bar code image is that size is 161 * 664 256 gray scale black and white images, and promptly the black picture element value is 0, and the white pixel value is 255.Under a predefine value situation, can embed 1200 bit informations; (a) original image; (b)-(e) provide the predefine value respectively and be 50,100,150,200 the watermarking images that comprises; (f) provide the experimental result that adopts 50,100,150 and 200 4 piece image gray levels simultaneously.

Claims (11)

1. digital information safety transmission and methods for using them is characterized in that based on the spatial domain digital watermarking algorithm of orthogonal template and the digital watermarking algorithm of two-value/many-valued discrete picture, and is carrier safe transmission digital information with the digital watermarking.

2. a kind of digital information safety transmission and methods for using them according to claim 1 is characterized in that a kind of spatial domain image watermark algorithm based on orthogonal template.Mutually orthogonal characteristic is satisfied in the design of orthogonal template, if p orthogonal template (p is the positive integer greater than 1): T is promptly arranged ₁, T ₂..., T _p, their conditions that should satisfy are T so ₁ T ₂ ... T _p, wherein represents quadrature.The watermark incorporation model is the additivity watermarking algorithm in addition, and wherein the calculating of scale factor s is calculated according to the permission distortion factor index of watermarking images automatically and finished; Three, the extraction of watermark is finished with coherent detection; Four, the method for many orthogonal templates is adopted in the embedding of many bit informations in image block.

3. a kind of digital information safety transmission and methods for using them according to claim 1, it is characterized in that utilizing spatial domain image watermark algorithm that non-compression video stream (as the video sequence that image pick-up card directly collects, AVI video flowing or the like) is carried out digital watermark processing based on orthogonal template.The telescopiny of watermark can be real-time, online, also can be non real-time, off-line.Be about to the image sequence that video sequence is decomposed into single frames, design the two-dimensional quadrature template then, according to watermark information, the application image watermarking algorithm, watermark information is embedded in each two field picture, watermark extracting also with the image watermark class of algorithms seemingly, directly extract the watermark information of every two field picture, thereby realize watermark embedding and leaching process video sequence.

4. a kind of digital information safety transmission and methods for using them according to claim 1 is characterized in that utilizing the spatial domain image watermark algorithm based on orthogonal template that compressed video data (as the MPEG-1/2 video flowing) is carried out digital watermark processing.Digital finger-print, characteristic information, copyright protection information or other control information are embedded in the key frame of video flowing as watermark, thereby in the video flowing of compressed format, protect special watermark information.

5. a kind of digital information safety transmission and methods for using them according to claim 1, it is characterized in that the voice data (as MP3 format) to compressed format and the voice data (as the WAV form) of non-compressed format, utilize spatial domain watermarking algorithm to carry out digital watermark processing based on orthogonal template.Orthogonal template is an one dimension, adopts the watermark incorporation model of stack and the watermark extracting model of coherent detection.

6. a kind of digital information safety transmission and methods for using them according to claim 1, the discrete digital watermarking model that it is characterized in that a kind of two-value/multivalue image, the i.e. discrete watermarking algorithm of the big capacity watermark information of embedding in two-value black and white or many-valued discrete picture (as one dimension or two-dimensional bar, E-seal, handwritten signature, form, literal, line segment or curve or the like) through overscanning.This algorithm is image block according to the structure of discrete picture with image division, and what determine the image block size according to picture structure, the distortion level that comprises watermarking images and need embed watermark information amount.Image block can be square, round dot or other version.

7. a kind of digital information safety transmission and methods for using them according to claim 1, it is characterized in that utilizing the digital watermarking algorithm of two-value/many-valued discrete picture that watermark data is embedded in two-value or the many-valued discrete picture, make the figure/image conversion of digital information be able on the discrete picture (as literal, form, the black and white image after the scanning, seal, bar code or the like) realize.Comprise special image block in the discrete picture, watermark information just is embedded in the image block.Block-shaped can be square, round dot or further feature figure.The watermark information amount that embeds is decided by the image gray levels of image block.The invisibility of watermark then depends on the pixel value that image block is modified.

8. according to the digital watermarking algorithm of claim 6 and 7 described two-value/many-valued discrete pictures, it is characterized in that with the large capacity digital watermark be carrier, watermark information is embedded in the bianry image or Continuous Gray Scale/coloured image, figure, form or literal of electronic document (as ADOBE PDF, MS-WORD, POSTSCRIPT and other e-file etc.).

9. according to the digital watermarking algorithm of claim 6 and 7 described two-value/many-valued discrete pictures, it is characterized in that to comprise the bar code applications of digital watermarking in various electronics such as certificate, ID card, trade mark, bill and electronics, paper document and paper media by discrete digital watermarking algorithm based on two-value and multivalue image, embed necessary digital watermark information in the bar code, differentiate and other effect as false proof, anti-piracy, copyright control, the true and false.

10. a kind of digital information safety transmission and methods for using them according to claim 1 is characterized in that proposing a kind of certificate false proof function that comprises digital watermarking bar code and biological information.By watermark information being embedded in the bar code based on the spatial domain image watermark algorithm of orthogonal template and the discrete digital watermarking algorithm of two-value and multivalue image, watermark information can be certificate information or its digital signature such as certificate holder name, date of issue, also can be the authentication information of various necessity such as the proprietary biological characteristic of certificate (as face picture, fingerprint, iris etc.) information.

11. the anti-counterfeit method of credentials that comprises digital watermarking bar code and biological information according to claim 10 is characterized in that extracting the digital digest (HASH value) of biological information and applies it on the certificate false proof.Biological information at first carries out discretization to be handled, and calculates its HASH value then, and the HASH value of biological information is embedded in the certificate with the watermark form.

Images