粒子滤波检测前跟踪中邻域窗口大小参数分析

Analysis of Neighborhood Window Size for Particle Filter Based Track-Before-Detect

在粒子滤波检测前跟踪中,为更好利用量测信息,每个粒子需考察其所影响的多个分辨单元的量测。该邻域窗口参数增大时,PFTBD性能较好,但是计算量也成倍增加。从量测信息的利用角度来看,粒子数目也是一个很重要的因素。以理论分析和仿真实验,研究了不同粒子数下该参数对PFTBD性能的影响。结果表明,选择较小的邻域参数和较大的粒子数时算法的效率较高。

For the better utilization of the measurement information in Particle Filter based Track- Before-Detect （PFTBD）,each particle needs to examine the effects of the measurements of multiple resolution cells affected by it. When this parameter of neighborhood window size increases, the performance of PFTBD is better, but the calculation will also be doubled. For the perspective of the use of measurement information,the number of particles is also an important factor. Through theoretical analysis and simulation, this paper studied the effects of this parameter on the performance using different particle number. Simulation results show that choosing a small neighborhood window size with large particle number, a higher efficiency can be achieved.