计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (22): 201-204.

• 图形、图像、模式识别 • 上一篇    下一篇

一种基于智能场景分析的2D转3D视频时分算法

王小毛,黄传河,赵俭辉,陈  俊   

  1. 武汉大学 计算机学院,武汉 430072
  • 出版日期:2012-08-01 发布日期:2012-08-06

Time-division algorithm for converting planar video into 3D video based on inteligent scene analysis

WANG Xiaomao, HUANG Chuanhe, ZHAO Jianhui, CHEN Jun   

  1. Computer School, Wuhan University, Wuhan 430072, China
  • Online:2012-08-01 Published:2012-08-06

摘要: 2D转3D视频的快速实现算法主要有时分法和位移法。对时分法的原理和缺陷进行了分析,在此基础上提出了一种新的基于场景特征的时分算法,它能更好地3D化平面视频。该算法利用了基于视频帧图像的颜色信息和运动矢量信息对视频进行自动分割的方法,根据视频段场景特征自适应选择左右眼最佳配对帧方法,以及通过视频段运动矢量信息来调整左右眼对应帧和补帧的方法来改进处理和立体化视频。对合成立体视频的立体显示实验分析证明改进方法有效地改善了合成立体视频的质量。

关键词: 2D转3D, 时分法, 位移法, 视频分割, 补帧

Abstract: The two main methods of fast converting 2D video into 3D video are Time-Division and Pixel-Dislocation. This paper, analysing the principle and defect of the traditonal Time-Division method, puts forward a new algorithm based on Inteligent scene analysis which can better convert planar video into 3D video. This new Time-Division method has used video automatically division  based on color character and motion vector information and eyes frames matching  algorithm and frames filling method to improve the 3D effect. Experimental results in real stereoscopic display test show that the new method effectively improves the quality of the transformed 3D video.

Key words: convert 2D into 3D, time-division, pixel-dislocation, video division, frames filling