QoS based SVC video adaptation in MPEG-21 framework

Abstract

Recently, usage environment for multimedia contents has been highly variable and heterogeneous. Considering these characteristics, scalable video coding (SVC) can be regarded as a good solution to cope with heterogeneity and variability. To serve optimal quality of service to user, method for quality control to extract the best quality of SVC video at the given constraints should be considered. In this paper, QoS based SVC video adaptation method in MPEG-21 framework is proposed. The main goal of this work is to find the best way for QoS based SVC video adaptation at a given bit-rate and support it in MPEG-21 DIA framework. For this purpose, perceptual quality preferences are considered as a QoS policy to guide bit-stream extraction process rather than PSNR or MSE. Perceptual quality preferences are the result of subjective test on the classified video scenes (action, crowd, dialog, scenery, and text&graphic). To support MPEG-21 DIA framework, SVC adaptation operators and effective way of describing those values in AQoS Description have been introduced. Quality decision algorithm has been structured to apply those DIA classification schemes to SVC bit-stream extraction. To verify the effectiveness of the proposed method, SVC video adaptation has been performed by the determined QoS policy in MPEG-21 framework. Test sequences have been selected to represent each class of QoS, and proper QoS policy has been applied to SVC bit-stream extraction process depending on sequence``s characteristic (Football sequence for action class, Harbour sequence for scenery, and Crew sequence for dialog class). The results (spatial, temporal, and SNR quality) of the extracted SVC video satisfying the given constraints (network bandwidth profile) shows the feasibility of the proposed method.