A dynamic frame-skipping video combiner for multipoint video conferencing

• DOI: 10.1109/iscas.2002.1010242
• Publication Date: 2003-06-25T20:42:54Z
• Authors: Kai-Tat Fung; Wan-Chi Siu; Yui-Lam Chan
• Keywords: Videoconferencing; Teleconference; Motion vector; Transcoding; Frame rate

This paper proposes a new architecture of video combiner for multipoint conferencing. The proposed first identifies the active and talking conference participants from multiple incoming bitstreams by computing their corresponding motion activities level audio signals. By reducing frame rates inactive sub-sequences, more bits are reallocated to sub-sequences such that higher obtained produce better visual perception as well smoother motion. We also propose low-complexity high-performance frame-skipping transcoder which is achieved direct summation DCT coefficients non-moving macroblocks an adaptive vector refinement compensated macroblocks. then used realize conferencing some results presented show improvement in performance due our architecture.