摘要
本文提出了一种二维视频到三维视频的转换方法。采用时域和空域相结合的方法,根据边缘信息,从图像序列中提取运动对象。对二维视频序列来说,其本身并不包含有视差信息,但是可以根据运动信息来估算视差。本文中采用基于三角形网格的浓密视差提取方法来估算。根据双目立体视觉原理和立体显示技术,我们将二维视频帧经过变换,得到伪三维立体视频帧,并合成视频序列。
The thesis is about the method of pseudo stereo video reconstruction based on two-dimensional video sequence. It adopts the methods of edge detection as well as moving information to segment the movement of the video according to information in time domain and space domain.According to the research on disparity calculatation, the method of extracting dense disparity based on the triangle lattice is adopted, although two-dimensional video sequence doesn＇t contain the disparity information, the disparity can be estimated based on the information of moving in the video sequence. According to the theory of binocular stereo vision and stereo display technology, two-dimensional video frames are changed into the pseudo threedimensional video frames, and they are reconstructed video sequence.
出处
《重庆电子工程职业学院学报》
2010年第1期78-81,共4页
Journal of Chongqing College of Electronic Engineering
关键词
运动估计
视频分割
视差
movement estimation
video segmentation
disparity
