DIGITAL WATERMARKING USING PRINCIPAL COMPONENT ANALYSIS AND ARNOLD CAT MAP SCRAMBLING TECHNIQUE

Abstract

In recent years, the advancement in the digital multimedia technologies has brought many facilities in reproduction, transmission and data manipulation. However this advancement has also brought the problem such as copyright issues. For copyright protection of digital multimedia, watermarking has been proposed. A watermark is embedded into the data which will indicate whether the content is copyrighted or not. Digital watermarking is a vast research area which is gradually growing. Over the years, researcher in Digital watermarking community has developed numerous new techniques for watermark embedding and detection. For the analysis of these techniques different attacks and counter-measures were being performed, encouraging for development of better techniques. Digital watermarking may be distinguished according to visibility (visible and invisible), robustness level (fragile, semi-fragile and robust), media type (audio, image, video) and the need for original data (blind, semi-blind, non-blind). The main scope of this thesis is to provide a good trade-off between the perceptual quality of the video after watermarking and the robustness of the video against various attacks. In the work, a method known as Arnold Cat Map is used to scramble the watermark image before embedding to provide some level of security to the watermarking algorithm. Watermark is embedded into the new sub space obtain from principle component analysis. Here in the thesis different semi-blind and non-blind techniques are performed. The methods were tested in Matlab and were analysed using Peak Signal to Noise Ratio (PSNR) and Normalised correlation for similarity measure. The satisfactory outcome of the technique can be seen in the watermarked video and the extracted watermark.