A new technique for real-time visual tracking of 3D objects

Abstract

A new technique for real-time estimation of the pose of a moving object using a video camera is presented. A Binary Space Partition (BSP) tree is adopted to represent the object geometry and efficient selection techniques are developed to find an optimal set of feature points of the object to be used for feature extraction. The extended Kalman filter is used to estimate the pose of the object directly from the image measurements. The performance of the proposed approach is experimentally tested in the case of an object moving in the visible space of one fixed camera.