A Stereo Vision Based Target Tracking and Obstacle Avoidance

摘要: Designing a visual tracking system to track an object is complex task because large amount of video data must be transmitted and processed in real time. In this study, stereo vision used acquire the 3D positions target, can achieved by applying CAMSHIFT algorithm, then apply fuzzy reasoning control steer mobile robot follow selected target avoid in-path obstacles. The adopted obstacle avoidance component based on Harris corner detection binocular imaging, which performs correspondence calculation. Therefore depth map created showing relative distances detected substantial features robot, provides information obstacles front wheeled robot. designed servo less sensitive lighting influences thus more efficiently. Experimental results showed that successfully finished target-following avoiding