Research On The Application Of Stereo Vision Based On Fast SIFT Algorithm With A Single Camera

Vision-based robot navigation technology is a navigating method developing in recent years. Compared with other traditional non-vision navigation systems, visual navigation is more excellent and effective. In this paper, visual telemetry is proposed for the mobile robot navigation. The study includes three parts: feature extraction, single stereo vision and environmental reconstruction. Details are as follows:A revised SIFT Algorithm has been used in the feature extraction. In addition, it has simplified the Algorithm itself, and thus improved the progress.A large number of accurate feature points have been used, which offers a geometrical restriction in the stereo-matching. With other conditions, this has made it possible to realize binocular stereo vision under single vision system.On the base of stereo-geometrics, the stereo vision system with a single camera has set up its mode of measurement. It has applied auto error revision in this progress, which enhances the accuracy of localized objectA visualized reconstruction of obstruction by using the two dimensional distance data has been realized under the applying conditions.The results of the tests have concluded the proposed single camera system can make an accurate space measurement of obstruction in the surroundings.