A Theoretical Framework for End-to-End Video Quality Prediction of MPEG-based Sequences

Abstract

This paper presents a novel theoretical framework for end-to-end video quality prediction of MPEG-based video sequences. The proposed framework encloses two discrete models: (i) A model for predicting the video quality of an encoded signal at a pre-encoding stage and (ii) A model for mapping QoS-sensitive network parameters (i.e. packet loss) to video quality degradation. The efficiency of both the discrete models is experimentally validated, proving by this way the accuracy of the proposed framework.