Development of a 3D Video Camera by using Polynocular Stereo

点击翻译按钮获取中文摘要

This research project is a successor of our previous research project, "3D image display system with motion parallax presentation capability", which was done from 1995 to 1996. It has developed a 3D photography system. In this project, we have tried to develop a 3D video camera system. The difference of the 3D photography and 3D video camera is the real time capability. Our polynocular stereo algorithm developed in the previous project needs a few minutes for the processing of one frame. As a 3D video camera, it is necessary to process 25〜30 frames in every second. In this project, we have succeeded to demonstrate the feasibility of our polynocular stereo algorithm for the video-rate processing. The major results developed in this project are as follows. We have developed a novel algorithm integration scheme to realize a video-rate high precision 3D video camera system. It integrates two stereo algorithms with and without occlusion handling mechanism. We also have developed a stereo core algorithm that fully utilizes the MMX architecture of a Pentium processor. It realizes the video-rate stereo processing on an ordinary PC. An experimental 3D video camera system has been developed.