中文摘要：

为解决多传感器探测下非机动群内目标精细跟踪的难题,深入分析群内目标回波特性,基于形状方位描述符,提出了一种新的多传感器群内目标精细跟踪算法,该算法利用形状方位描述符建立可唯一表示群内各目标组成图形的形状矢量,然后基于落入群内各目标相关波门内的量测集合,以群内各目标可能关联量测构成的所有图形为对象,通过形状相似程度和量测与目标状态一步预测值的空间距离建立相似度模型,同时利用选主站的思想实现冗余图像的剔除,最后利用量测集合与对应的权值集合,基于粒子滤波技术实现群内各目标的状态更新。经仿真数据验证,与传统多传感器多目标跟踪算法中性能优越的基于数据压缩的集中式多传感器多假设算法相比,该算法在跟踪精度、实时性、有效跟踪率三个方面的性能明显优越,能很好地满足群内目标精细跟踪的实际工程需求。

英文摘要：

To solve the refined tracking problem of the non-maneuvering group targets under multi-sensor detections,the intensive analysis on the characteristics of the group target measurements is made.Based on the shape and azimuth descriptor,a new multi-sensor group targets refined tracking algorithm is proposed.Firstly,a shape vector is established by use of the shape and azimuth descriptor,which can represent the graph of each target in the group.Then,based on the measurement set falling within the gate of each target,making all graphes generated by the associated measurements of each target in the group as the object,the similarity model is established through the shape similarity and the distance between the measurement and the target state one-step prediction value.At the same time,redundant images are removed by using the idea of selecting the master station.Finally,each group target status updating is achieved based on particle filter by using measurement set and the corresponding weight set.Compared with the centralized multi-sensor multi-hypothesis algorithms based on data compression,this algorithm has obvious advantages in the three aspects of tracking accuracy,real-time performance and effective tracking rate.It can meet the engineering requirement of the refined tracking of the group targets under multi-sensor detections.