基于随机有限集理论的多扩展目标跟踪关键技术研究

项目名称： 基于随机有限集理论的多扩展目标跟踪关键技术研究

项目编号： No.61305017

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 杨金龙

作者单位： 江南大学

项目金额： 25万元

中文摘要： 随着现代雷达等探测设备分辨率的不断提高，数目未知且时变的多扩展目标跟踪已经成为目标跟踪领域的研究热点和难点问题。针对多扩展目标跟踪，由于存在多扩展目标量测产生率不确定问题；不规则形状多扩展目标量测集难以划分问题；多扩展目标紧邻、遮挡或分裂时，易导致目标被错估计或漏估计问题；多扩展目标平行和交叉运动时航迹难以正确关联问题等，本项目以随机有限集和概率假设密度为理论支撑，拟开展以下研究内容：(1)设计未知量测产生率估计方法；(2)融合量测产生率，设计自适应量测集划分方法，以解决扩展目标紧邻及形状不规则的量测集划分问题；(3)对不同的扩展目标分别建立形状模型,针对扩展目标紧邻、遮挡和分裂时设计新的状态更新策略，以提高扩展目标状态和形状的估计精度；(4)针对多扩展目标的紧邻、平行和交叉运动时航迹关联问题，充分挖掘目标的有用信息，设计有效的扩展目标航迹管理方法，实现多扩展目标的连续跟踪。

中文关键词： 扩展目标跟踪；概率假设密度；量测划分；不规则形状估计；航迹关联

英文摘要： With the development of high resolution sensors, the multiple extended target tracking techniques for an unknown and time-varying number of targets have become a hot topic and an extremely difficult problem in the target tracking field. For the multiple extended target tracking, the following problems are very common: the measurement rates are unknown; it is hard to plot the measurement sets of the multiple extended targets with irregular shapes; the targets are inclined to be wrongly estimated or omitted when the targets are close, shielded or splitted; and it is hard to associate the trajectories correctly when the targets move in parallel or cross each other, etc. To solve these problems, we will carry out the following research works based on the theories of the random finite set and probability hypothesis density: (1) estimating the unknown measurement rates by designing an effective method; (2) designing an adaptive method for plotting the measurement sets by integrating the measurement rate estimates; (3) constucting the shape model for each extended target and designing the update strategies for different stages of the close, shielded or splited targets, which can improve the accuracy of state and shape estimates; (4) designing the track maintenance algorithm by digging up the important information of th

英文关键词： Extended target tracking；probability hypothesis density；measurement partition；irregular shape estimation；track association