Indian Journal of Science and Technology

Abstract

Objectives: The present work proposes a secure and robust video watermarking algorithm to detect the camcorder piracy using Discrete Cosine Transform (DCT) and Schur Decomposition. Methods/Analysis: Separate watermark for each video used to improve security. The watermark is further secured using scrambling by Arnold transform and then dividing it into subparts. These scrambled watermark chunks embedded into audio sequence and video frames using robust technique DCT and Schur decomposition. The security has been improved with randomness in frame selection block selection. Recovered information from extracted watermark can be used to authenticate the owner. Findings: Experimental results show that the proposed algorithm is not only robust to various attacks but also highly secure. Perceptual quality of watermarked frame has been compared with the original frame. The Peak Signal to Noise Ratio (PSNR) is 49.21 dB for MB.mp4 video and 48.88 dB for GT.mp4 video file after embedding watermark. Robustness of watermarked video has been tested against various signal processing, video processing attacks and camcorder attack. Extracted watermark is compared with original watermark and Normalized cross correlation (NK) value is calculated, which is coming nearly 1. The security has been strengthened with randomness in frame selection, block selection, Scrambling and division of the watermark. A Watermark is embedded in audio sequence and video frames. Novelty/Improvement: Position of pirate is estimated by comparing original video frames with watermarked video frames. Mean Absolute Error is 2.8 cm. 

Keywords: Arnold Transform, DCT, Schur, SVD, SCD