【作者】 蒋定定；

【导师】 韩勇；

【作者基本信息】 中国海洋大学 ， 地图学与地理信息系统， 2010， 硕士

【摘要】 本文分析、总结了数据关联的诸多方法,从应用角度讨论了信息融合的功能模型和融合方法,以及目标跟踪技术中的一些基本要素。论文以科研项目“装甲车辆多传感器信息实时处理系统”作为研究背景,阐明了多目标跟踪所采用的主要算法,并着重分析了在几个关键问题上所采用的处理方案。首先,文中介绍了数据关联算法中的最近邻算法和数据关联快速算法,并进行了仿真实验和结果对比分析,验证了二者的优缺点。在前人研究的顺序多传感器联合概率数据关联算法基础之上,基于数据关联快速算法,文中提出了一种新的关联算法—顺序数据关联快速算法,该算法的基本原理是：凡是落入跟踪门内的点迹,都认为是所跟踪目标的一个有效回波,只是每个有效回波的互联概率不同,而互联概率是根据有效回波的正态分布计算得到,最后利用第一次滤波结果和观测数据进行第二次滤波以减少误差。同时对该算法和数据关联快速算法进行了仿真及误差对比分析,验证了顺序数据关联快速算法的有效性。其次,针对战场目标跟踪中雷达获得的目标角度误差大的问题,引入了高精度的红外方位探测信息进行关联,用以修正融合航迹点的方位角,该算法充分利用了红外与雷达的互补性,经仿真实验,降低了目标航迹的方位角和俯仰角跟踪误差,取得了较好的实验效果。更多还原

【Abstract】 In this paper, many methods of data association are analyzed and summarized. Functional model and fusion methods are discussed from applied angle. In addition, basic elements in targets tracking technology are introduced. Furthermore, based on the project which systemic study for information processed real-time of armored vehicles is introduced, the general framework and working flow are explained and some methods of several key steps have a detailed analysis in this paper.Firstly, the nearest neighbor for data association (NNDA) and fast algorithm for data association (FAFDA) are introduced in this paper. At the same time, emulation experiment and analysis of emulation result are conducted, which validates merits and defects between NNDA and FAFDA. This paper presents a novel method of data association based on fast algorithm for data association and multiple sensors joint probability data association, which is called multiple sensors fast algorithm for data association (MSFAFDA). The basic theory of MSFAFDA is that all the data in gate are regarded as valid data. But associated probabilities of the valid data are different. The second data are calculated according to the first data and observational data. At the same time, emulation result and error contrast and analysis between NNDA and MSFAFDA validate it.Secondly, in allusion to question which the precision of targets gained by radar is low, azimuth gained by infrared sensor with high precision is introduced to improve azimuth precision of targets fused. This algorithm makes good use of merits between radar and infrared sensor. Emulation experiment validates that azimuth error of targets fused is reduced rapidly.更多还原