%0 Journal Article
%T 基于JNDD边界划分的立体视频深度编码<br>Stereoscopic Video Depth Coding Based on JNDD Boundary Segmentation
%A 王来花
%A 侯春萍
%A 朱涛
%A 王宝亮
%A 阎维青
%J 天津大学学报（自然科学与工程技术版）
%D 2016
%R 10.11784/tdxbz201506070
%X 针对编码后的深度图中边界区域失真导致的虚拟视图质量下降的问题, 利用基于恰可察觉深度差异(JNDD)的边界提取方法, 将深度图划分为尖锐的边界区域和平滑的非边界区域, 并对划分后的区域分配不同的量化参数值进行深度编码．实验结果表明：本文提出的方法能够很好地保留深度图的边缘信息, 有效地减轻虚拟视图绘制的失真; 同时, 绘制的虚拟视图能够保持较高的峰值信噪比(PSNR)和结构相似性测量参数(SSIM)值．<br>The distorted edge regions in depth map caused by depth coding will result in the quality of virtual view reduced. Depth map is divided into sharped boundary regions and smoothed non-boundary areas through the method that boundary segmentation is based on the just noticed depth difference(JNDD) and assign the different quantization parameters to the segmented regions，respectively. The experimental results show that the proposed method can well retain the edge information of depth map and effectively reduce the distortion of rendered virtual view. The rendered virtual view can also maintain high values of peak signal to noise ratio(PSNR)and structural similarity measurement parameters(SSIM)
%K 虚拟视点合成图
%K 深度失真
%K 立体视频编码&lt
%K br&gt
%K rendered virtual view
%K depth distortion
%K stereoscopic video coding
%U http://journals.tju.edu.cn/zrb/oa/darticle.aspx?type=view&id=201609013