Video Coding Based on Bitplane and GFA Modeling

The H.264/AVC technology emerged in 2003 has achieved at average a bitrate gain of 40-50% over the previous low bitrate coding technologies, namely H.263, H.263+, H.263++ and H.26L but at the same time has seemingly pushed the block-based motion compensation video coding to the limit in rate-distortion performance. In this paper, we present a novel video coding scheme based on the generalized finite automata (GFA) modeling of video sequences in bitplane wavelet domain. In the proposed scheme, a video sequence is represented in Groups of Pictures (GOPs) as an overall binary image by bitplane modeling the significant coefficients of the video sequence within subbands. The inter-frame, inter-level and inter-bitplane similarities inhabited in the binary image are then optimally explored, leading to a compact GFA representation of the bitplane video sequence. Finally, all the transitions in the GFA representation are entropy encoded into a scalable bitstream. The proposed scheme significantly outperforms the H.26X series coding schemes in rate-distortion performance and can achieve bitrate ranges at 4-5 Kbps and 15-18 Kbps for QCIF 10 Hz and QCIF 30 Hz sequences, respectively