Effective and Efficient Design Space Exploration for Heterogeneous Microprocessor Systems-on-Chip
Chao Wang, Peng Chen, Xi Li, Xuda Zhou, Xuehai Zhou, Nadia Nedjah · 2015
Digital video watermarking is the art of hiding ownership data in video such that an originator can prove his/her authenticity whenever required. H.264, also known as MPEG-4 Part 10 advanced video codec, is one of the latest standards, developed jointly by ITU-T video coding expert group and ISO/IEC & Motion Picture Expert Group (MPEG) (Richardson 2004). It has a high data rate and also provides better picture quality. The compression efficiency is two/four times than MPEG-2/ MPEG-4 coders, respectively (Wiegand et al. 2003). H.264 is a useful standard for cable TV and HDTV for Digital Video Broadcasting Handheld (DVB-H) systems with small screens (Dengpan et al. 2012). Real-time video watermarking is desirable for applications such as copyright protection, video authentication, video fingerprinting, copy control, and broadcast monitoring. Most of the work done in video watermarking till now was focused on uncompressed domain watermarking whereas existing electronic devices like digital camera and mobile phones provide output in one of the video compression standards. In uncompressed domain watermarking, the video is decompressed in order to embed the watermark, and then watermark embedding algorithm is run on it, which is used to embed the watermark in original video that is again recompressed to produce a secure compressed stream. For real-time applications, software watermarking is not an ideal choice as it introduces substantial delay in generating the watermark-embedded video. All software-based algorithms can be targeted on digital signal processor (DSP) implementation, but this approach is not suitable in case of realtime applications. Real-time video watermark embedding can be achieved where a custom integrated circuit (IC) resides in electronic appliances, which embeds the watermark at the time when video is captured. The embedding unit should be a part of the encoder for real-time watermark embedding. The design of a watermarking chip involves trade-offs among various performance parameters such as power, area, and real-time performance (Mathai et al. 2003b). In the present work, an algorithm with pipeline and parallel architecture is proposed, which has better speed suited for real-time requirements (Kougianos et al. 2009).