XION PRO를 사용한 물체 인식 및 추적 시스템

In this paper, We propose the surroundings detection system that can be applied to a exploration robot, life-saving robot. Using the ASUS’s XTION PRO Live, the system is configured in three steps to recognize the walls and obstacles, perceive objects and to track the selected object. First of all, the surface is recognized by a quadric equation obtained by experiments. BLOB(Binary large object) Labeling should be applied to depth image to perceive objects and obtain three-dimensional coordinates of the objects. Next step is to select one of the labeled objects as a target. For tracking of the target, CamShift provided by OpenCV was applied on RGB image since CamShift is based on Hue values.