A no-reference video quality metric by using inter-frame encoding characters

Macro-block (MB) in DCT domain which contains the temporal and spatial information of video is a good video quality estimator in both encoding and transmission phase. Furthermore, MBs encoded by block-based coding standards can be obtained easily in bit-stream with a low time-complexity algorithm. Even so, this parameter is rarely used in video quality estimation under lossy network. So the major contribution of this paper can be concluded as it imports information from MBs of inter-frame encoded frames into video quality prediction and proposes a universal no-reference video quality metric over wireless networks which is named as Inter-frame Video Quality Metric (IEVQM). In detail, IEVQM analyze the impacts of both encoding and channel conditions to the video quality degradation by using motion vector and residual error from received P-frame and/or B-frame. Through extensive experiments and human subjective tests, we show that IEVQM demonstrate high correlation with Mean Opinion Score (MOS) for unknown video sequences both in distributed loss and burst loss situation (higher than 0.81).