一种由单视频图像序列获取多视差图的方法研究

A new method of obtaining multi-parallax images from a single video

通过对二维视频中帧间物体与摄像机位置关系和现有的二维视频转三维的方法分析,提出了一种由单视频图像获取多视差图的方法.通过利用单视频连续帧与多相机系统拍摄同一场景获得的多视差图之间的相似性,提出的方法可实现从单视频连续帧中获取所需要的视差图,并对其可行性作了理论和实验上的比较分析,实验证明了作者提出的方法在某些特定环境中的可操作性.

According to the analysis of the location relations between an object and camera in different frames in a 2D video and the 2D to 3D conversion methods used, this article proposes a new kind of method of obtaining multi-parallax images from a single video image serials. This method utilizes the disparity information between frames and then combine these multi-parallax images to form a combined image which is used for 3D auto-stereoscopic display. The experimental shows that the new method is feasible in some specific environment.