The most popular and widely deployed media communication applications are likely to be those that are accessible over the Internet, both through the currently predominant wired access and through emerging wireless access channels. This requires the design of compression techniques that are robust to channel impairments (variations in bandwidth and delay, packet losses, bit errors, etc). Our work in this area concentrates on: (i) Long-term memory motion compensation, where the motion compensation gain can be significantly increased by extending the motion search range with multiple decoded frames as reference frames, instead of using just one decoded frame as in conventional motion compensation, (ii) Multiple Description Layered Coding (MDLC), where the advantages of layered coding and multiple description coding techniques are combined to provide robust video communication by adapting to the varying channel conditions on the fly, (iii) Wyner-Ziv Scalability (WGS), where based on the Wyner-Ziv framework, the proposed coder can support embedded representation and high coding efficiency by using the high quality version of the previous frame as side information in the enhancementlayer coding of the current frame.

Tech Summary
NSF Report (Year 7)
NSF Report (Year 8)

Poster

Several optimal and suboptimal joint temporal-spatial rate control solutions to video coding and streaming will be presented. Based on that, Rate- Distortion models are applied to reduce the algorithm complexity and processing delay time, such that the proposed approaches can be used for real- time video streaming. Several video clips and selected frame pictures will be shown to demonstrate the advantages of the proposed methods comparing to the MPEG- 4 standard.

Tech Summary