Two Dimensional Edge and Corner Model based Object Tracking using Particle Filter

In this paper we suggest a method for a parametric object tracking using Particle Filter theory. In the method we solve the two problems faced in object tracking; object segmentation difficulty comes from illumination variation and cluttered background, and motion complexity of object itself. To settle the first problem we use the feature of distinct corners and edges, and for the second problem we model the DOF of the object motion in 2D and 3D with mathematical description. To observe the similarity of a hypothesis model in the image; the image and transformed feature model with a 3D transform matrix, we adopt chamfer matching method. From the experiment using an object of A4 size, this algorithm is very robust for 2D motion of rigid objects. C. Oh, M. Z. Islam, C. Lee, Two Dimensional Edge and Corner Model based Object Tracking using Particle Filter, 15th Japan-Korea Joint Workshop on Frontiers of Computer Vision (FCV 2009), pp. 223-228, Feb. 2009