Spatial quality index based rate perceptual-distortion optimization for video coding

• We have a deep study on Spatial Quality Index (SQI).
• An effective and low complexity perceptual-distortion is defined as signal distortion tuned by contrast and motion masking effect factors.
• An SQI-based rate perceptual-distortion optimization algorithm is proposed to improve subject video quality.

Spatial Quality Index (SQI) is a recently proposed video quality assessment metric that can predict video quality much close to subjective judgments. Since current video coding still has much redundancy in the sense of visual perception, in this paper, we incorporate SQI into video coding to further improve compression ratio without visual quality loss. Firstly, contributions of different human visual system (HVS) properties used in SQI are analyzed. Then two most important HVS properties, i.e. contrast masking effect (CME) and motion masking effect (MME), are extracted to measure perceptual-distortion Dp in video coding. Finally, based on Dp, a rate perceptual-distortion optimization (RpDO) algorithm is presented by adopting a suitable Lagrange multiplier from previous study. Experimental results show that, RpDO can averagely achieve 14% bitrate reduction when compared to HM14.0 under the same visual quality. At the same time, there is no significant change in the encoding time.