Optimized Particle Filter for Tracking of a Moving Video Object

Abstract

Recently, the particle filter (PF) is used to track video object by estimating its position in image frame, but it suffers some problems: degeneracy phenomenon and sample impoverishment. In this paper, a new optimized bootstrap particle filter (OBPF) is introduced to solve the problem of particle filter on video object tracking for rigid motion. Where adaptive Chemostic step , bacteria use in adaptive bacteria foraging based on particle swarm optimization algorithm (ABF-PSO), is tackled to improve predict step in PF. Moreover, particle swarm optimizations (PSO) is used in resample step of OBPF. The comparison among OBPF, particle filter based on bacteria foraging optimization (PF-BFO) and particle filter based particle swarm optimization (PF-PSO) on video object tracking is presented in this paper by using Matlab program. The results show that OPF method presents outstanding performance versus PF-PSO and PF-BFO. It has the best velocity and the best accuracy, so it is suitable for real time tracking