View-dependency video coding for asymmetric resolution stereoscopic views
View-dependency video coding for asymmetric resolution stereoscopic views
复制标题
DOI：
10.1117/1.3156016
复制
发表时间：
2009-07
影响因子：
1.3
通讯作者：
Seanae Park;D. Sim
中科院分区：
文献类型：
--
作者：
Seanae Park;D. Sim
我们提出了一种可用于商用立体显示器的非对称分辨率立体视频编解码器。目前，大多数立体视频显示器被设计成不仅显示立体视频，而且还显示传统的单视点视频，以便使它们与单视点视频服务兼容。在这些系统中，立体模式的每个输入视频的分辨率是单视模式的输入分辨率的一半。我们提出了一种立体视频编解码器，通过对全分辨率左图像和水平下采样的右图像进行编码来同时支持单视和立体服务。然而，由于左图像的分辨率不同于对应的右图像的分辨率，因此不可能在两个视点之间采用传统的视差估计。因此，我们提出了一种基于一种新的视差估计/补偿方法的立体视频编解码器，该方法利用了左参考图像的所有信息而不需要下采样。与几种传统的立体编码算法相比，本文提出的立体编解码器获得了0.5-0.8分贝的编码增益。
We propose an asymmetric-resolution stereoscopic video codec that can be used for commercially available stereo displays. Currently, most stereo video displays are designed for displaying not only stereo video but also conventional single-view video in order to make them compatible to single-view video services. In these systems, the resolution of each input video for stereo mode is half the input resolution for a single-view mode. We propose a stereoscopic video codec to support both single-view and stereo-view services by encoding full-resolution left images and horizontally downsampled right images. However, it is not possible to employ the conventional disparity estimation between the two views because the resolution of the left image is different from that of the corresponding right image. So we propose a stereoscopic video codec based on a new disparity estimation/compensation method that makes use of all the information of the left reference image without downsampling. We achieved a 0.5- to 0.8-dB coding gain with the proposed stereoscopic codec, compared with several conventional stereoscopic coding algorithms.