A Novel Robust Contour Tracking Algorithm

Abstract

Under the conditions of fasting moving, shape changing and partial occlusion of target, Now mostly tracking algorithms are difficult to get accurate contour of target, for these problems, a novel contour tracking algorithm based on particle filter and fast level set is proposed. Firstly, the particle filter algorithm is adopted to estimate moving target’s boundary contour. Then, according to the nearest neighbor decision method, a new the velocity function of fast level set is found, the strongpoint of the function is that the tracking algorithm is fit for target and background changing. Finally, the contour tracking is realized by evolving the zero level set curve using fast level set algorithm which is proposed in this paper. Experiments for representative image sequences show that this algorithm can track the rigid and non-rigid target contour under the complex environments. The result indicates that this algorithm is robust and accurate compared with other tracking algorithms.