Moving object tracking - a parametric edge tracking approach

Abstract

In this paper, an edge based tracking algorithm is proposed. Our algorithm makes efficient use of edge-segment on the Canny edge map by utilizing the edge structure in the moving object region. Curvature-based features are used for moving edge registration. We use the maximum curvature correspondences between two edge segments then the 2D affine transformation computes their movement by solving a system of linear equations. The registration error is then minimized. A Kalman Filter is used to track each individual edge segments. Segments are clustered using a k-mean algorithm. Finally, a group motion tracker is used for tracking moving object from each cluster. Experiments show that our edge-segment based tracking algorithm can track moving objects efficiently under varying illumination conditions.