Temporal segmentation of video sequences for content-based coding

Abstract

We present an overview of the state of the art of segmentation based video coding aimed at low-bit rate application. The most significant contribution is the algorithm for the segmentation of videoconference image sequences for content-based coding. The segmentation is based on motion estimation through the computation of the optical flow field and motion segmentation by applying a graph-theoretical clustering. The algorithm will be added to the video codecs of the OpenH323 project which is based on H.323 ITU-T Recommendation. This visual collaboration environment for Desktop, composed of terminals, gatekeepers and MCU (Multipoint Control Unit), has already been implemented and tested. It is particularly convenient for academic implementation, the software is freely distributed, and its code can be modified according to the needs of the user. However, the first tests using the H.261 codec showed a video delay of 3 seconds, confirming the need to optimize the video codec.