Scalable predictive coding using coset codes.

摘要

The proliferation of digital image capture devices and the Internet have fueled a tremendous amount of research in the fields of multimedia signal processing and multimedia communication over the last decade. This dissertation addresses the problem of compression and communication of multimedia content over unreliable channels. We propose an alternative to the conventional predictive coding of video based on the formulation of predictive coding as a variant of the Wyner-Ziv side-information problem in information theory. Through this formulation we propose the design of two video coding algorithms that encode video signals in a bandwidth scalable and error-resilient manner, respectively. Empirical evaluation confirms that the proposed coding approach compresses video signals with minimal source coding loss over conventional approaches, while simultaneously providing flexibility and resilience in the context of communication over unreliable channels.