Optimal Video Adaptation and Skimming Using a Utility-Based Framework

Video adaptation uses various spatio-temporal adaptation techniques to transform video content in order to satisfy diverse resource constraints and user preferences. To explore systematic solutions, we present a new conceptual framework to model content, adaptation processes, utility, resources, and relations among them. The key to the framework is definition of the notion of utility and the formulation of the problem as one of constrained utility optimization. In this paper, we use the framework to form a unified view towards problems and solutions we have developed in several media adaptation projects, ranging from video skimming, transcoding, to adaptive streaming. We discuss several important open issues, including ambiguity in adaptation specification, representation of resource and utility distributions in high-dimensional spaces, modeling and classification of resource/utility, and efficient solution search strategies under multiple complex constraints.