| The rapid growth of digital imagery, the increasingly easy access to digital media, and the increasingly powerful tools available for manipulating digital media have made media security a very important issue. Digital watermarks have been proposed recently as the means for intellectual property right protection of multimedia data. Digital watermarking is a process of embedding information (or signature) directly into the media data by making small modifications to them. With the detection/extraction of the signature from the watermarked media data, it has been claimed that digital watermarks can be used to identify the rightful owner, the intended recipients, as well as the authenticity of a media data. We shall discuss watermarking of image data in this paper, although the general idea presented here is also applicable to other forms of multimedia data. The watermarks can be perceptually visible or invisible. We focus on invisible watermarks in this paper. In general, there are two most common requirements of invisible watermarks. Perception: The watermarks should be perceptually invisible. Robustness: They should not interfere with the media being protected. They should also be robust to common signal processing and intentional attacks. Particularly, the watermark should still be detectable/extractable even after common signal processing operations have been applied to the watermarked image. These include filtering, compression, geometric transformation, cropping, etc. In addition, the watermark should be resistant to intentional attacks that attempt to remove the watermarks. It should be noted that the two basic requirements, imperceptibility and robustness to signal processing, conflict with each other. In fact, one of the fundamental problem in digital watermarking is to achieve a good balance between these two requirements. In this paper a new watermarking scheme is proposed based on the perception and robustness. In addition we also develop the non-blind watermarking scheme and the blind watermarking scheme respectively. We focus our attention on these topics: ◆HVS: The Human Vision System is very important to the watermarking scheme. First we describe the Human Vision System form physiology. Then we define the JND in DCT and DWT domain based on Waston model. By this way we can improve the quality of watermarked image. ◆Mathematics foundation: Because the watermarking algorithm is indeed a mathematical problem. On the other hand we mainly use wavelet transform in this domain we insert watermark. So we review some mathematical foundation, then we can deal the problem easily in the consequence parts. ◆In this section, a new watermarking scheme is proposed based on the human visual system (HVS), discrete wavelet transform (DWT) and the orthogonal vector. This method can exactly extract the watermark. Especially, it can resist the signal enhancement operation. In addition, the proof of the performance and security is given. The embedding procedure is: First we set two orthogonal vectors S0 and S1 like below: S0 = [0 1 0 0 1 1 1 1 0 0 0 1 1 1 0 0];IIIS1 = [1 0 1 1 0 0 0 0 1 1 1 0 0 0 1 1]; The coefficients are modified: W (i ,j) = ??? HH11(( ii,,jj))++ aallpphhaa**JJNNDD((ii,,jj))**SS01((( ( i i --11))??44++jj)) ,,iiff WW′′== 10 ;???i,j=1....4 The non-blind detection procedure: PH0 = ABS(S UM(( H L?HL)??S0)) PH 1= ABS(S UM(( H L-HL)??S1)) W~ ′( i,j)=0 ;if PH0?PH1>0 W~ ′( i,j)=1 ; if PH1?PH0>0 The blind detection procedure: We first define a function Part. It can divide the S into two parts S1 , S 2,and ???? ∑? ∑????argmin i∈S 1 Si j∈S2Sj. Part() { 2T = sum(S); While S ≠Φ{ For each Si ∈S, find argmin (a bs(S i ? T)) S = S?Si; US1 = S1Si; T = sum(S 2) ?Si For each Sj ∈S, find argmin (a bs(S j ? T)) S = S?Sj; US 2 = S2Sj; T = sum(S 2) ?Sj } } By Part function the orthogonal vector S0, S1 are determined dynamically. Then S0, S1 are transmittedsecurely using a key. In the detection procedure, S0, S1 are obtained using the key. Then the watermark can be extracted exactly just like the non-blind watermarking detection procedure. The analysis of security: For blind watermarking scheme of this algorithm, the most important point is to obtain the embedding position. By transmission S0, S1, we can know the right position essentially. Now we give the proof of security. Proof: Supposing a violator knows the algorithm exactly and don't know the key, then he can't obtain S0, S1 In the embedding procedure, the equation holds: ∑∑∈∈=i S1 jS2S i Sj After the embedding the equation holds: ∑∑∈∈≠i S1 jS2S i Sj The observed signal isS~ . Supposing a violator knows the the function Part exactly and don't know the key, so he can't obtain S0, S1. It means the embedding position is uncertain. The violator then can't detect the watermark exactly using our algorithm. End of proof. ◆We have performed experiments. 18 commonly used attacks are adopted to test the robustness of our method. We also... |
PDF Full Text Request