Video Copyright Protection Using Chaotic Maps and Singular Value Decomposition in Wavelet Domain

Abstract

This paper attempts to offer a new method for video copyright protection using Singular Value Decomposition (SVD) and chaotic maps in the discrete wavelet domain. To provide integrity to the watermark image, the singular values of watermark images are permuted using a hyper chaotic map and an authentication key. Based on the number of scene changes in the video, the permuted watermark image is partitioned into a number of blocks. Then the frame difference method is applied for selecting the key frames in each scene. In the embedding process, each frame in the group is transformed to single level DWT. The principal component (PC) of the watermark is embedded into the LL subband of each scene in the specific key frames and the corresponding permuted watermark PCs are added to the SVs of the scene followed by inverse SVD to obtain the watermarked scene. The suitable scaling factor is selected by a new mechanism in tune with rates of changes of PSNR and NCC. The proposed algorithm is robust against video processing attacks with better integrity of watermark data with lesser processing time. The proposed method also resolves the false-positive problem by embedding the principal component of the permuted watermark.