Efficient Video Transport over Lossy Networks

Nowadays, packet video is an important application of the Internet. Unfortunately the capacity of the Internet is still very heterogeneous because it connects high bandwidth ATM networks as well as low bandwidth ISDN dial in lines. The MPEG-2 and MPEG-4 video compression standards provide efficient video encoding for high and low bandwidth media streams. In particular they include two paradigms which make those standards suitable for the transmission of video via heterogeneous networks. Both support layered video streams and MPEG-4 additionally allows the independent coding of video objects. In this paper we discuss those two paradigms, give an overview of the MPEG video compression standards and describe transport protocols for Real Time Media transport over lossy networks. Furthermore, we propose a real-time segmentation approach for extracting video objects in teleteaching scenarios