The Discussion On Multi-bits Watermarking And Fingerprint Watermark Based On Multi-bits Methods

With the development of digital technology and Internet, while digital product shows its convenience, its security also challenges the digital work owner, because the software pirate can also apply this convenience at the same time and invade owner's rights. Thus it is already extremely urgent to take a variety of measures to protect the digital product and to penalize those pirates for the right infringement, but besides the law and management methods similar to the traditional work copyright protection, the technical protection should be provided according to the special characteristics of digital product. The research of digital watermark technology is precisely developing rapidly under such kind of application request.The digital watermark used in the intellectual property rights protection has appeared the important commercial value, but the digital watermark system still had the obvious insufficiency in the theory and the realization method. And the human body fingerprint characteristic has fine performance such as uniqueness, stability, inviolability and so on. Its development is most mature in the biology characteristic. In view of the insufficiency of the digital watermark system, this paper proposes the mentality that introduces the fingerprint recognition into the watermark system, applies the digital watermark technology and the fingerprint recognition technology synthetically, and both of them have the widespread application prospect.Along with the development of technology, the digital watermark algorithm also unceasingly improves and enhances. But it can not completely protect the copyright depending on the technology only. The copyright protection must rely on the management and the technical realization protocol as well as a set of complete secure watermark system. At present, the malicious attack of watermark is mainly on the protocol layer, this request that we must seek the corresponding measure from the protocol layer. Because the human body biology characteristic recognition technology is a convenient, fast, and accurate status recognition method, it can solve some difficulty problem in the watermark technology, such as watermark key security transmission, uniqueness and the unidirectional and so on. Introducing the biology characteristic recognition technology into the digital watermark protocol can confirm the both sides status of the sale and the consumer, ensure the sale's legitimacy, avoid other people pretending the sale or the consumer to carry on transaction this kind of intermediate attacked, constitute the safe digital watermark protocol.Most watermark algorithm of previously uses "zero" bit watermark, but now as a result of application, it requires that extract out a little meaningful watermark information when checking, such as the text, the picture, the fingerprint and so on, used to prove the proprietary rights of product or work, this need to realize with multi-bits watermarking technology, therefore, the multi-bits watermarking has the broader application prospect. Multi-bits watermarking has the well application in the broadcast surveillance, the copy protection and the access control, the pirate aspect tracing and the copy right protection and so on.The insert way of multi-bits watermarking has the direct information encoding and the multi-symbolic encoding these two ways, the direct information encoding is a very good method when the bit number is small, but is not realistic when the number of bit is bigger, therefore the paper selects the multi-symbolic encoding method to realize multi-bits watermarking. If using the code modulation directly, the obtained information vector will have the bad code division. Therefore this paper uses the error correcting coding technology. In the field of the algorithm, the paper adopts the space division multiplexing which the LSB and the HSB unifies, the DWT frequency division multiplexing and the space division and the DCT frequency division union multiplexing these three algorithms. This paper takes the experiment with 4 bits watermark, first carries on the hamming code to the watermark to correct error, and adds the redundant position, then causes it to become 8 bits watermark signals. Under to three algorithms briefings as follows:(1) In the space division multiplexing, the watermarks signal is embedded by the LSB and the HSB algorithm. In the LSB algorithm we use the pseudo-random sequence to decide the position of watermark signal embedded to. In the HSB algorithm we use the compensated means to reduce the influence on the image quality by the change of highest element position of image;(2) In the DWT frequency division multiplexing, because the watermark embed in the wavelet low frequency coefficients is not easy to lose, but it is easy to cause the image quality change, and the watermark embed in the wavelet high frequency coefficients is not easy to realize by person's vision, but the watermark is easy to lose. Therefore this paper unifies the different characteristics of the high and the low frequency coefficients, it carries on the compromise consideration, it adopts the algorithm that embeds more watermark information in the high frequency coefficients and embeds the less watermark information in the low frequency coefficients;(3) In the space division and the DCT frequency division union multiplexing, first it divides the primitive image into 9 images those have the same size using the interlacing sampling piecemeal method, which according to the space division method, then it chooses 8 images in these 9 images to embed watermark, it carries on the DCT transformation to the image block and embeds the watermark into the DCT intermediate frequency coefficient.When appraising watermark algorithm, this paper introduces the concept that Just Noticeable Difference (JND) for objective appraise the transparency of algorithm, besides using the Peak Signal-to-Noise Ratio. With the aid of the JND that appraise standard, it introduces a variableβ, it expresses normalized processing's value that the pixel number that the gray minus in the image D more than in the primitive image's JND value, the image D is the absolutely minus of primitive image I and the image I 'which have embed watermark. Theβis used to appraise the transparency of algorithm.The biological recognition based on the fingerprint characteristic information technology is a biology recognition technology, the development of it is earliest, the application of it is most widespread and the technology of it is most mature, therefore this paper chooses the fingerprint recognition this kind of recognition technology to realize the biological characteristic watermark. It uses the space and frequency unifies multiplexing algorithm, realizes fingerprint characteristic information embedding with two kinds of method:Method 1: It takes multi-bits watermark signal with fingerprint characteristic data. First, this method modulates the fingerprint characteristic data to watermark signal that satisfies the algorithm, the fingerprint characteristic data is transformed to the binary sequence, and it uses BCH(15,7) error correction code that error correction ability is 2, then it carries on the compression and encryption processing to it. Then it divides the primitive image into 16×16 sizes blocks, embeds 1 bit watermark information in each piecemeal image.Method 2: It takes multi-bits watermark signal with fingerprint characteristic point. This method divides the fingerprint image and primitive image both into 16 16 sizes blocks, if there is a characteristic point in fingerprint image block, then it embeds the different watermark in the corresponding primitive image block according to the characteristic point type.×During the embed process, with the Gaussian distribution pseudo-random sequence loading watermark information, it uses the DCT algorithm, embeds the watermark into intermediate frequency coefficient. When the digital work has the copyright dispute, it can withdraw the fingerprint characteristic data which embed into advance, carry on the contrast with the copyright owner's fingerprint, then it checks them whether matches, determines its copyright owner, and thus it has protected the copyright owner's legitimate rights and interests. After the test, the watermark embed algorithm proved to have the very good transparency and robustness, it can satisfy the need of many kinds of situations. The method uses the multi-bits technical realization the significance watermark, with the pseudo-random sequence loading each bit watermark information, it will be better than the way that direct embed the significant the watermark in the robust aspect.This method carries on the dual protection with the biology characteristic recognition technology and the digital watermark technology to the digital work, it can cause the digital work owner and the legitimate user's rights and interests obtain a better safeguard. Its original characteristic and the superiority, must be powerful promote the development of digital watermark theory that protection intellectual property rights, must be promote the application of technical. Its industry prospect is considerable. The achievement promotion will produce the considerable economy and the social efficiency. The prospect of digital watermark is optimistic, and our research work also will last forever. Standing on predecessor's shoulder, we base ourselves on the present and focus in the future.