A Novel Motion Estimation Algorithm for very Low Bit Rate Video Coding

Abstract

This paper presents a novel video coding algorithm for very low bit rate applications. It gets rid of the problem of transmitting of large amounts of region shape information, because this can be obtained by segmenting the reconstructed picture at both ends. A novel Coiner detection and tracking approach is used for robust motion estimation. This motion estimation approach is computationally much more efficient than the previously suggested approaches which use optimization methods. Corner points are used to represent a region, and estimation of their motion, through tracking, is used to characterize the motion of the region. Computer simulations, using DCT coding to code the error have shown estimated bit-rates of 15.3 kbps. The results indicate objective and subjective performance of the coder is comparable to existing low bit rate algorithms with a much lower computational load and noise susceptibility.