(An) optimal adaptation framework for video transcoding in universal multimedia access

Abstract

Many techniques exist for adapting videos to satisfy heterogeneous resource conditions or user preferences. However, selections of appropriate adaptations among various feasible choices are often ad hoc. First, to provide a systematic solution, we present a general conceptual framework that models adaptation, resource, utility, and relations among them. It allows formulation of various adaptation problems as resource-constrained utility maximization. Second, a simple method to estimate the distortion caused by the DCT coefficient dropping for rate adaptation is presented. It is expected that the method can be employed to effectively allocate the distortion among frames in coefficient-drop transcoding applications. Third, we propose a content-based statistical approach to facilitate prediction of a utility function for real-time transcoding of live videos. Finally, we present a description tool to support utility-based adaptation for terminal and network quality of service (QoS).