Unified Platform-Independent Airborne Networking Architecture for Video Compression; Final rept. Mar-Dec 2007

摘要

It is highly desirable to have a video coding framework that can flexibly distribute coding complexity between the video encoder and decoder. However, today's video compression techniques impose a rigid computational complexity distribution between the video encoder and decoder. Specifically, video decoders are very simple but encoders are computationally complex as they need to carry out motion estimation and mode decision. Distributed video coding has shown great potentials in enabling flexible complexity distribution between the video encoder and decoder. Previous work on distributed video coding has studied two extreme operating points of complexity distribution: (1) Complex encoder with light-weight decoder to achieve maximum compression efficiency and (2) Light encoder and complex decoder to achieve robustness against transmission packet drops (at the cost of reduced compression efficiency). In this project, our main goal is to explore the middle ground between these two extreme points. In particular, we study the effect of doing coarse motion search at the encoder.