Based On Integer Wavelet Transform Digital Watermarking Technology

With the rapid development of the digital multimedia distribution over the Internet, digital watermarking, a novel technique for copyright protection, credibility and integrity authentication of digital media, has attracted more and more attention. Since 1990s, It has been introduced to digital watermarking technology in information hiding area , and the research on digital watermarking has getting more and more in-depth and broad. At present, digital watermarking tech has been widely used in copyright protection,video on demand,e-business,Multimedia Message Service etc. It has become one of rot topic in new multimedia information security community.In this thesis, a robust watermarking algorithm was focused on static image. Quantization Index Modulation (QIM) algorithm based on lifting wavelet transform and a double-eye QIM watermarking based on Multiresolution Subband Decomposition (MSD) - Independent Component Analysis (ICA) were proposed.Firstly, the theory of digital watermarking was introduced. The coefficient transformed by Discrete Wavelet was floating-point number, which is not suit for Numeral Computing. According to that, an improved QIM algorithm was developed based on Lifting Wavelet Transform (LWT), which was also implemented by MATLAB. Last it has the conclusion that the algorithm based on LWT has strong robustness when experience attack, and watermarking can be extracted after that.Now multi-scale analysis and independent component analysis technology were used in feature extraction broadly. It's also applicant to embed and extract watermarking. A Double-eye watermarking algorithm based on MSD -ICA was proposed. According to Human Visual System, watermarking was embedded and extracted in insensitive eye area, and the robustness of watermarking was analyzed.Contributions of this dissertation were depicted as follows:1. A robust digital watermarking algorithm based on QIM and LWT was developed.This algorithm accelerated Numeral Computing. The running time was reduced andmemory was saved. This achievement has been published in "journal of computer science".2. In this thesis, a double-eye watermarking method based on the fusion of MSD -ICA for watermarking extraction was proposed. Firstly, original image was decomposed by LWT, and the watermarking was embedded into the middle-frequency coefficients. Secondly, watermarking was extracted from red gray image and blue gray image respectively; it has been applicant ICA technology in order to remove noise. Lastly, when the watermarking was detected, the combination of the two methods held their advantages.All the processed of the algorithms presented in this thesis, including embedding,extracting and attacking were realized by MATLAB7.0. Experimentation proved thatthe watermark can also be extracted or detected if the digital watermarking algorithm isattacked by adding noise, filtering, and JPEG compression.