基于深度信息的多视点视频的高效表示和编码研究

项目名称： 基于深度信息的多视点视频的高效表示和编码研究

项目编号： No.61202240

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 计算机科学学科

项目作者： 刘美琴

作者单位： 北京交通大学

项目金额： 22万元

中文摘要： 下一代电视（三维电视）先进的显示技术给人的视觉带来强烈的冲击和震撼。 随着显示技术的发展和观众享受自然3D世界的需求，创建更加多元化和更加逼真的三维显示技术成为迫切需要解决的问题。三维视频是经典二维视频的扩展，最明显的特征就是在传统二维视频的基础上增加深度信息，由于双目视差的原理，用户可获得具有深度感的立体图像，增强了视觉的现实感和逼真感。三维视频的应用场景十分广阔，但是三维视频数据量非常庞大，这对三维视频的高效表示和压缩提出了严峻的考验。在三维视频表示方面，基于深度图的场景表示方法，因场景遮挡及深度数据不连续将会导致中间图像出现较大范围的空洞，对图像重建质量造成严重的影响；在多视点视频编码方面，现有算法大多只去除时间冗余，并没有完全去除视角间的冗余，极大地影响了多视角编码的性能。本项目将从这两方面出发，在研究深度图像的基础上，提出能有效提高重建图像质量的算法和能高效压缩多视点视频的算法。

中文关键词： 三维电视；多视点视频编码；深度图；；

英文摘要： As the next generation of television, three-dimensional television (3DTV) has entertained us with its advance in display, just like a feast for eyes and is enkindling various new applications. With the development of display technologies and requirement of audience to enjoy the natural 3D world, how to strive to create realistic 3D impressions of natural 3D scenes seems to be more urgent. The major challenge is that natural 3D scenes representing results in tremendous amount of data because 3D adds information of depth which is different from 2D videos. In other words, the key to success in 3DTV is efficient compression since there are too much data needed to be transmitted under limited bandwidth. Since the representation and coding of three-dimensional video has direct on successive transmission, synthesis and display, it becomes crucially important and receives a wide publicity. Various 3D display technologies use various kinds of data representation formats, adopting different coding methods. But their objective remains the same, which is to encode 3D video by removing temporal and spatial redundancy. This project focuses on coding algorithms of various representation formats. By studying depth image seriously, we give pay attention to improve quality of 3D scenes and better the compression performance by st

英文关键词： Three-dimension television；Multi-view video coding；depth map；；