Visual Saliency and Distortion Weighting Based Video Quality Assessment

Video quality assessment (VQA) is very important in many video processing applications. For example, the rate-distortion (RD) optimization in video coding needs an efficient distortion metric to assess the RD cost of candidate coding parameters. However, most existing metrics employ little visual perceptual information, or some are too complex to meet real-time requirement. In this paper we propose a new model called saliency and distortion weighted structural similarity index with temporal pooling strategy (SDTW-SSIM). In the proposed model, spatial and temporal saliency is obtained from the referenced video. Besides, a distortion weighting map is employed to give a full description of visual attention. To better present the perceptual properties of videos, both frame and sequence level saliency features are taken into account. Experimental results show that, compared with state-of-the-art methods, the proposed method performs well on both computational efficiency and assessment accuracy.