Research on quality assessment metric based on H.264/AVC bitstream

Abstract

No-reference(NR) video quality metrics are more practical in real-time applications compared to full-reference(FR) metrics. This contribution proposed a No-reference video quality assessment metric based on H.264/AVC bitstream through extracting features from the H.264/AVC encoded bitstream. After the extraction of the features which are very important for video quality assessment, we use Partial Least Squares Regression(PLSR) to calculate the weights of them. Then a quality prediction model is also proposed. During the experiments, the results show that our NR metric has low computing complexity. Finally, compared to subjective assessment, we find that there is a high correlation between quality prediction and the actual quality of 0.95.