摘要
粒子滤波是适用于非线性非高斯系统下目标跟踪的强有力工具.MiroSot足球机器人系统可以作为研究机动目标跟踪问题的平台.对此,在分析MiroSot系统目标特征的基础上,提出一种基于目标特征约束的均值漂移粒子滤波算法,利用约束和优化的思想提高粒子的质量并减少其数量.对比实验表明,该方法有效地克服了传统粒子滤波的计算量和粒子退化问题,保证了多机动目标跟踪的准确性和实时性.
The particle filter can track targets in the situation of non-linearity and non-Gauss.MiroSot robot soccer system is used as the paltform.By analysing the features of targets in this system,a feature-restricted and mean shift-based particle filter algorithm is proposed,in which the thought of optimization from restriction is used to improve the quality of particles and decrease their quantity.The contrast experiments show that the proposed algorithm can overcome the disadvantage of traditional particle filter and track maneuver targets aecuratelly and in real time.
出处
《控制与决策》
EI
CSCD
北大核心
2010年第1期149-152,共4页
Control and Decision
基金
国家863计划项目(2006AA04Z212)
河北省教育厅自然科学研究项目(Z2008473)