基于联合特征的边缘粒子滤波目标跟踪算法研究  

Marginalized particle filter for combined feature target-tracking

作　　者：孟军英[1] 刘教民[2] 韩明[2] 

机构地区：[1]石家庄学院计算机科学与工程学院,石家庄050035 [2]燕山大学信息科学与工程学院,河北秦皇岛066004

出　　处：《计算机应用研究》2015年第6期1906-1911,1916,共7页Application Research of Computers

基　　金：河北省自然科学基金资助项目(F2012208004);河北省高等学校科学技术研究重点项目(ZD20132013)

摘　　要：针对采用单一特征建立的动态空间模型与实际系统差距较大,从而使估计误差增加的问题,通过将系统的状态参数引入颜色特征模型中,与颜色特征参数一起构成系统状态空间向量,提出了一种联合颜色状态特征的优化目标跟踪算法。应用Rao-Blackwellization算法思想,由Kalman线性滤波方法解析处理线性的颜色特征转移和更新过程;而目标位置参数采用粒子滤波进行估计,提高了视频目标跟踪的精度和实时性。通过与其他相似算法的比较实验结果可以看出,算法在环境亮度发生变化、目标遮挡等情况下,仍能够保持较高的跟踪精度,既提高了跟踪系统的鲁棒性,又保证了算法的实时性,优于传统的单一特征视频跟踪算法。In order to improve accuracy of system model and reduce evaluated error resulting, this paper proposed a margina- lized particle filter algorithm with joint characteristic. Depart from color appearance parameters, the location parameter intro- duced into system model as another system state vector. Furthermore, it split the state into two parts：the location parameters, which were estimated stochastically by particle filter, and the color appearance parameters, which were estimated analytically. Both parts were jointly tracked using Rao-Blackwellization technology. Rao-Blackwellization algorithm reduced the dimension of mixed linear/nonlinear system and realizing a higher tracking precision and lower computation complexity. Comparison experi- ment results demonstrate that the proposed algorithm is able to automatically update when ambient brightness and speed of tar- get changes or the target is sheltered. The algorithm is more robustness and real-timeness than the traditional color-based parti- cle filter tracking algorithm.

关 键 词：目标跟踪 边缘粒子滤波 核概率密度估计 卡尔曼滤波算法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] TP301.6[自动化与计算机技术—计算机科学与技术]