Model-based Robust Video Watermarking

Please enable javascript in your browser.

Bakhshi Golestani, Hossein | 2013

623 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 44108 (05)
University: Sharif University of Technology
Department: Electrical Engineering
Advisor(s): Ghaemmaghami, Shahrokh; Gholampour, Iman; Ghanbari, Mohammad
Abstract:
Nowadays, H.264 is a popular format to data transfer. The growing use of video signals and importance of its ownership in the digital world, requires careful study of robust video watermarking. Distorting the watermark through recompression is a common attack to robust video watermarking schemes. It is aimed in this thesis to design a recompression resistant watermarking scheme and to address real-time extraction of the watermark from both compressed video bitstream and decoded video. In this thesis, we employ an objective quality metric, which is highly correlated with human perception, as a cost function in image/video watermarking optimization. We show that by embedding in the selected blocks, the imperceptibility could be improved significantly.
A two-layer scalable data partitioning (DP) based watermarking is also proposed to enhance robustness against intense attacks. In this approach, the watermark is divided to more and less important parts and the more important part is embedded stonger than less important parts. Simulation results show that in intense attacks, the DP based method outperforms normal method which the watermark is embedded uniformly
Keywords:
Robust Video Watermarking ; H.264 Video Codec ; Video Recompression ; Data Partitioning

No TOC