Non-redundant frame identification and keyframe selection in DWT-PCA domain for authentication of Video

This study is intended to protect video data and watermark from unauthorised access. The proposed methodology accentuates two new algorithms, namely structural similarity index metric–absolute difference metric (SSIM-AMD) based non-redundant frame identification (NRFI) and entropy–AMD based keyframe selection (KFS) to reduce the challenges posed by traditional discrete wavelet transform–singular value decomposition. Traditional techniques embed the entire watermark to all existing frames in the video, which is cumbersome and time-consuming. In this methodology, NRFI algorithm is applied to segregate the redundant and non-redundant frames to specific database. The KFS algorithm is used to identify suitable keyframes. DWT is applied into keyframes, which decomposes the frames into subbands. The middle band is selected for embedding. The principal component of watermark image block is embedded into identified keyframes in the video. The chaotic map is adapted to reorder the watermark block for improving the authentication level of the watermarking. The ant colony optimization (ACO) technique is adapted to select the suitable scaling factor for watermarking process. The principal component analysis technique is employed for avoiding false-positive attacks. Experimental results show the proposed methodology can withstand image processing, video processing, false-positive attacks and produces good results in terms of perceptual quality and robustness.