4 November 2016 Video coding for 3D-HEVC based on saliency information
Author Affiliations +
Abstract
As an extension of High Efficiency Video Coding ( HEVC), 3D-HEVC has been widely researched under the impetus of the new generation coding standard in recent years. Compared with H.264/AVC, its compression efficiency is doubled while keeping the same video quality. However, its higher encoding complexity and longer encoding time are not negligible. To reduce the computational complexity and guarantee the subjective quality of virtual views, this paper presents a novel video coding method for 3D-HEVC based on the saliency informat ion which is an important part of Human Visual System (HVS). First of all, the relationship between the current coding unit and its adjacent units is used to adjust the maximum depth of each largest coding unit (LCU) and determine the SKIP mode reasonably. Then, according to the saliency informat ion of each frame image, the texture and its corresponding depth map will be divided into three regions, that is, salient area, middle area and non-salient area. Afterwards, d ifferent quantization parameters will be assigned to different regions to conduct low complexity coding. Finally, the compressed video will generate new view point videos through the renderer tool. As shown in our experiments, the proposed method saves more bit rate than other approaches and achieves up to highest 38% encoding time reduction without subjective quality loss in compression or rendering.
© (2016) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Fang Yu, Ping An, Chao Yang, Zhixiang You, Liquan Shen, "Video coding for 3D-HEVC based on saliency information", Proc. SPIE 10026, Real-time Photonic Measurements, Data Management, and Processing II, 100261E (4 November 2016); doi: 10.1117/12.2245389; https://doi.org/10.1117/12.2245389
PROCEEDINGS
10 PAGES


SHARE
Back to Top