Mixing and playback of JPEG compressed packet videos

This paper addresses the issue of video mixing, which is to simultaneously display multiple motion videos received from multiple independent sources across a packet switched network. A straightforward approach, which performs mixing at pixel domain, is first analyzed and considered undesirable because of the requirements of large amount of expensive memory and decompression capacity. A new technique, which performs mixing directly with JPEG compressed data, is then introduced. An efficient structure based on this new technique is presented which needs only a frame worth of buffering and decompression regardless of the number of video windows and their sizes. The paper also addresses the packet video playback issue, and proposes two simple heuristics.