摘要
提出一种基于灰色系统理论和粒子滤波的目标跟踪算法,自适应调整搜索范围,并采用交叉熵理论来度量目标模型与粒子确定区域特征模型之间的相似度。先用粒子滤波算法对运动目标状态进行估计,确定目标中心位置,利用历史目标位置状态序列,通过灰色系统理论对下一帧目标状态进行预测;然后对重采样后的粒子的位置和粒子的搜索范围进行修正,采用交叉熵理论来衡量目标与粒子确定区域的特征模型之间的相似度。仿真结果表明,相比传统的粒子滤波算法,新算法具有更好的鲁棒性和跟踪精度。
This paper proposes a gray system theory and particle filter-based target tracking algorithm.It adjusts the search scope adaptively,and uses cross-entropy theory to measure the similarity between the model of target and the model of particle determined regional characteristics.First,the particle filter algorithm is employed to estimate the state of the moving targets and to determine the centre of the target,the state of next frame target is predicted using states sequence of historical target location and through gray system theory;then the position of the resampled particles and the search scope of the particles are amended,the cross-entropy theory is used to measure the similarity between the target and the feature model of the particles determined region.Simulation results show that compared with traditional particle filter,the new algorithm has better robustness and tracking accuracy.
出处
《计算机应用与软件》
CSCD
北大核心
2013年第4期131-134,共4页
Computer Applications and Software
基金
辽宁省教育厅科学研究项目(L2010202)
关键词
目标跟踪
粒子滤波
灰色系统
交叉熵
Target tracking Particle filter Gray system Cross-entropy
