运动目标跟踪的粒子滤波算法  

Particle Filtering Algorithm for Moving Target Tracking

作　　者：于勇[1] 车燕芳[1] YU Yong CHE Yan-fang(The 723 Institute of CSIC,Yangzhou 225001 ,China)

机构地区：[1]中国船舶重工集团公司第七二三研究所,江苏扬州225001

出　　处：《舰船电子对抗》2017年第4期46-49,58,共5页Shipboard Electronic Countermeasure

摘　　要：提出一种基于对数极坐标变换的改进粒子滤波算法,以解决传统粒子滤波算法中由于红外成像目标信息量少所导致的目标跟踪稳定性低、易于发散的问题。该算法在粒子滤波的理论框架下,利用跟踪区域的对数极坐标变换作为目标描述,建立了一种稳健的系统观测模型,有效突出了目标信息,减弱了目标的旋转和尺度变化对跟踪算法的影响。该算法结合了粒子滤波运动估计的鲁棒性以及对数极坐标变换目标描述的稳健性,可在复杂环境中实现对目标的稳定跟踪。An improved particle filtering algorithm based on log-polar transform （LPT） is presen- ted,which can solve the problems that target tracking stability is low and easy to be diverged due to little infrared imaging target information in the traditional particle filtering algorithms. Based on the theory of particle filtering, the algorithm uses the LPT of tracking area as target description, sets up a stable system observation model, which effectively highlights the target information, weakens the influence of target rotation and scale transformation on tracking algorithm. The algo- rithm combines the robustness of particle filtering movement estimation and the solidity of LPT target description, which can realize the stable tracking to target in complex environment.

关 键 词：运动目标跟踪 粒子滤波 对数极坐标变换 灰度直方图 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]