Adaptive Tracking Algorithm Based on Particle Grouped Optimization

LIU Zhentao,WANG Zhaoying,LIU Weiqun

DOI: https://doi.org/10.3969/j.issn.1671-637x.2012.02.008

2012-01-01

Electronics Optics & Control

Abstract:To address the problem of particle degradation in particle filter algorithm,an adaptive tracking algorithm based on particle grouped optimization was proposed.The conventional particle filter algorithm was improved by establishing a feature histogram and adopting particle resampling strategy.The conspicuousness and similarity of target and background were taken into consideration to set up a ratio relation for selecting the feature that could furthest differentiate the target from the background.The feature district amount was weighted by the discrimination.Through analyzing particle space distribution,a novel resampling strategy was proposed for adjusting particle amount and relative position adaptively by duplication,linear combination and elimination,which could optimize the particle performance.The effectiveness of the proposed algorithm was demonstrated by simulation experiments.