Research Of Quality Scalable Video Coding On HEVC

The introduction of the next-generation video standard called High-Efficiency Video Coding (HEVC) was finalized by the Joint Collaborative Team on Video Coding in January2013. Based on the suitable combination of the previous tools and new designs,50%bitrate savings can be provided approximately for equivalent perceptual quality relative to the performance of prior standards. In order to adapt to the diversity of networks and terminals, a quality scalable video coding theme is proposed to exceed the request of fast-developing video service. It is a research of major significance for video transmission in heterogeneous internet.In this paper, we propose scalable themes on both HEVC and three-dimensional High Efficiency Video Coding (3D-HEVC). Firstly, a quality scalable scenario on HEVC is proposed, and three inter-layer prediction algorithms are adopted to eliminate redundancy. Meanwhile, two-loop control structure and single-loop solution were employed to diminish the error and reduce complexity respectively. Secondly, we propose quality scalable extension design on three-dimensional video compression using HEVC based on the different scalable solution. An inter-layer texture prediction is extended into the proposed scalable scenario for texture views and depth maps by enabling a multi-loop decoder solution. A novel inter-layer distortionless prediction tool is added because of the smooth texture characteristic in depth maps to reduce complexity in the base layer. To further improve the coding performance, the prediction tool can be also used to utilize the coded prediction quadtree and texture data from corresponding distortionless blocks.In the scenario of quality scalable video coding based on HEVC, compared with the Single Layer, about19.86%Y-Bjontegaard Delta Rate (Y-BD-Rate) increase is measured. When compared to H.264/SVC and Simulcast,47.98%Y-BD-Rate saving and20.50%Y-BD-Rate reduction can be achieved. In the scenario of scalable video coding on3D-HEVC, the average37.2%Y-BD-Rate decrease of the can be gained in the inter-layer texture prediction compared to the simulcast for synth peak signal-to-noise ratio versus total bitrate. The inter-layer distortionless prediction has a38.9%decrease compared with the simulcast for synth PSNR versus total bitrate. Significant rate savings confirm that the proposed methods achieve better performance.