基于灰色预测模型和粒子滤波的视觉目标跟踪算法  

Visual tracking algorithm based on grey prediction model and particle filter

作　　者：朱明清[1] 王智灵[1] 陈宗海[1] 

机构地区：[1]中国科学技术大学自动化系,合肥230027

出　　处：《控制与决策》2012年第1期53-57,共5页Control and Decision

基　　金：国家自然科学基金项目(61075073);高等学校博士学科点专项科研基金项目(20093402110014)

摘　　要：结合灰色预测模型和粒子滤波,提出一种新的视觉目标跟踪算法.由于粒子滤波未考虑先验信息对建议分布产生的指导作用,不能很好地逼近后验概率分布,对此,采用历史状态估计序列作为先验信息,建立该序列的灰色预测模型来预测产生建议分布.与粒子滤波、卡尔曼粒子滤波和无迹粒子滤波进行对比实验,结果表明所提出的算法在视觉目标跟踪中具有更好的性能.In this paper, a visual tracking algorithm is proposed by combining particle filter with grey prediction model. Particle filter does not take into account the guidance of historical prior on the generation of proposal distribution, so that it can not approximate posterior density well; Therefore, the history of state estimation sequence is utilized as prior information to set up grey prediction model for predicting and generating proposal distribution. Through the comparison to particle filter, Kalman particle filter and unscented particle filter, the proposed algorithm exhibits better performance in visual tracking.

关 键 词：粒子滤波器 灰色预测模型 目标跟踪 建议分布 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]