PRISM: Video Coding with Flexible Complexity Distribution Between Encoder and Decoder (PRISM)
Kannan Ramchandran, Jiajun Wang and Chuohao Yeo
Air Force Research Laboratory and National Science Foundation
PRISM (Power-efficient, Robust, hIgh-compression, Syndrome-based Multimedia coding) [1,2], is a video coding paradigm based on the principles of distributed compression from multi-terminal information theory (often abbreviated as distributed video coding). More precisely, PRISM is based on the framework of lossy source coding with side-information (Wyner-Ziv theorem) with important modifications related to motion modeling in video. PRISM represents a major departure from conventional video coding architectures based on motion-compensated predictive coding (e.g., the MPEG-x, H.26x families) aiming to remove some of their architectural limitations. The key architectural innovations of PRISM are its inbuilt robustness to "drift" between encoder and decoder, and the potential of flexible distribution of computational complexity between encoder and decoder.
In our previous work, we have demonstrated that the distributed video coding paradigm can be applied to the two end points on the complexity spectrum, namely a maximally-complex encoder with a minimally complex decoder to achieve high compression efficiency [3,4], and a minimally complex encoder with a maximally-complex decoder to achieve high robustness to transmission packet drops [1,2]. Currently, we would like to explore the middle ground of complexity distribution while preserving the robustness of the compressed bit stream against transmission packet drops. Namely, we would like to explore how increasing the encoder complexity through very coarse motion search can help in:
- Reducing decoder complexity;
- Improving compression efficiency of the codec when there are not packet drops; and
- Preserving robustness of the bit stream when there are transmission packet drops.
To do this, we will redesign the classification process or even the classifier itself from the original system to incorporate the new information generated by a coarse motion search at the encoder.
- R. Puri and K. Ramchandran, "PRISM: A New Robust Video Coding Architecture Based on Distributed Compression Principles," Proc. Allerton Conference, October 2002.
- R. Puri, A. Majumdar, and K. Ramchandran, "PRISM: A Video Coding Paradigm with Motion-Compensated Prediction at the Decoder," IEEE Transactions on Image Processing, 2007.
- J. Wang, V. Prabhakaran, and K. Ramchandran, "Syndrome-based Robust Video Transmission over Networks with Bursty Losses," Proc. International Conference on Image Processing (ICIP), October 2006.
- S. Milani, J. Wang, and K. Ramchandran, "Achieving H.264-like Compression Efficiency with Distributed Video Coding," Proc. Visual Communications and Image Processing (VCIP), January 2007.