粒子滤波理论、方法及其在多目标跟踪中的应用

本文梳理了粒子滤波理论基本内容、发展脉络和最新研究进展, 特别是对其在多目标跟踪应用中的一系列难点问题与主流解决思路进行了详细分析和报道. 常规粒子滤波研究重点主要围绕重要性采样函数、计算效率、权值退化/样本匮乏和复杂系统建模展开. 作为一类复杂估计问题,多目标跟踪一方面需要准确的目标新生/消亡与演变、虚警/漏检等建模技术, 另一方面需要多传感器信息融合、航迹管理等复杂决策方法.暨有限集统计学应用于多目标跟踪后,粒子 滤波进入一个新的发展阶段---随机集粒子滤波.基于不同的背景假设,可以构建不同近似形式的随机集贝 叶斯滤波器并采用粒子滤波实现.但机动目标、未知场景、多目标航迹管理以及跟踪性能评价等仍是多 目标粒子滤波的研究难点和重点.

Particle Filtering: Theory,Approach, and Application for Multitarget Tracking

This paper reviews the theory and state-of-the-art developments of the particle filter with emphasis on the remaining challenges and corresponding solutions in the context of multitarget tracking. The research focuses of the general particle filter lie on importance proposal, computing efficiency, weight degeneracy, sample impoverishment, and complicated system modelling. Multi-target tracking involves a class of complex dynamic estimation problems that require both accurate models for target birth, death and evolution, false alarms and miss-detections, and efficient decision-making strategies regarding multi-sensor data fusion and track management. Specifically, with the introduction of finite set statistics to multi-target tracking, recent years have seen the burgeoning development of a new generation of particle filters, which is referred to as the random set particle filter in this paper. Based on different scenario assumptions, different approximate forms of random set Bayesian filters can be established and implemented by the particle filter. However, manoeuvring target, unknown scenario, track management and tracker performance assessment remain key challenges for the multi-target tracking particle filter.