Object Tracking Based on Particle Filter and Scale Invariant Feature Transform

Abstract

Particle filter is a popular stochastic tracker for object tracking. In this paper, we proposed a novel algorithm for object tracking based on particle filter and Scale Invariant Feature Transform (SIFT). The result of SIFT matching does not adopt to reweight the particles as previous methods, we adopts a hybrid schema to supplement the particle distribution of traditional factor sampling with importance sampling. Experiments show that the proposed algorithm yields a more robust tracking result.