Hand Tracking by Particle Filtering with Elite Particles Mean Shift

Abstract

This paper presents an improved particle filtering for hand tracking in a video sequence in real time. In this method, we incorporate mean shift process of elite particles into conventional particle filtering for hand tacking which aims to improve the tracking efficiency. The elite particles are chosen from the particles with higher weights based on observation model. The proposed method can significantly improve the tracking performance, as well as reducing the computational cost compared to conventional particle filtering. Experiments result of the hand tracking show the effectiveness of proposed method.