SCALABLE GENERATION AND CODING OF MOTION VECTORS FOR HIGHLY SCALABLE VIDEO CODING

One of the challenges of achieving highly scalable video coding (SVC) is the estimation and coding of motion vectors. In conventional video coders motion vectors are generated and lossless coded in a non-scalable manner. In SVC, motion vectors are estimated for the highest spatial resolution and they are scaled down at the decoder for lower spatial resolution accordingly. In such a case, the motion vectors cause a fixed cost irrespective of the decoding spatial resolution. In this paper we propose a simple, yet efficient, framework for scalable generation and scalable coding of motion information, for highly scalable video coding. This framework is based on the spatial resolution (scale) dependent precision limiting of motion vectors. The results show substantial improvements of spatial resolution scalability performance at low bit rates.