作者: Yue Li , Gaobo Yang , Yapei Zhu , Xiangling Ding , Rongrong Gong
DOI: 10.1145/3267128
关键词:
摘要: As a 3D extension to the High Efficiency Video Coding (HEVC) standard, 3D-HEVC was developed improve coding efficiency of multiview videos. It inherits prediction modes from HEVC, yet both Motion Estimation (ME) and Disparity (DE) are required for dependent views coding. This improves at cost huge computational costs. In this article, an early Merge mode decision approach is proposed texture depth maps in based on priori posterior probability models. First, model established by exploiting hierarchical interview correlations those previously encoded blocks. Second, built using Coded Block Flag (CBF) current block. Finally, joint adopted terminate Experimental results show that saves 45.2% 30.6% encoding time average while maintaining negligible loss efficiency, respectively.