SCALABLE CONTEXT-BASED MOTION VECTOR CODING FOR VIDEO COMPRESSION