超宽带雷达人体目标检测与跟踪

发布时间：2018-03-14 06:18

  本文选题：超宽带雷达　切入点：人体目标　出处：《国防科学技术大学》2014年硕士论文　论文类型：学位论文

【摘要】：短距离人体跟踪在安全领域应用上非常重要,比如机场安检、穿墙侦察恐怖分子、废墟中抢救受困人员。超宽带(Ultra-Wideband,UWB)雷达良好的距离分辨力和穿透能力,使其非常适合短距离人体跟踪。另一方面,工作环境的特殊性对超宽带雷达在目标检测、鉴别和跟踪方面提出了更高的要求。为此,本文将超宽带雷达对人体目标的检测和跟踪作为研究的重点。论文首先介绍了超宽带雷达的系统组成、信号体制及信号处理流程,详细分析了超宽带雷达常用的四种信号,并主要介绍了耦合对齐,背景相消,以及多径效应抑制三个问题。其中,耦合对齐能够提高背景相消抑制耦合强杂波的性能,指数加权背景相消则能够对径向和非径向运动的目标都有较好的动目标指示性能,利用时间窗可将多径和杂波从目标回波中分离出来。动目标指示后进行的目标检测,主要目的就是判断目标有无。传统的恒虚警检测(Constant False Alarm Rate,CFAR)方法对单个人体目标检测时,为使目标信息得到很好的保留,往往需要设置相对较高的虚警概率,但同时背景杂波也会随之增多;而在对多个相距很近,甚至交叉、重叠时的人体目标进行检测时,会出现严重的目标遮蔽现象。为此,本文提出将通常用于图像处理的CLEAN算法用于超宽带雷达人体目标检测中,相比CFAR算法,CLEAN算法对不同运动状态下的单个、多个人体目标均有较好的检测性能,且能够有效的抑制杂波、多径和目标遮蔽、自遮蔽现象,并能够很好的保留目标的信息,提取出人体多个散射点并记录下每个散射点的到达时延。本文基于人体多散射点的回波模型,利用CLEAN算法提取出人体多个散射点的量测信息后,结合最近邻数据关联算法和联合概率数据关联算法,分别实现了对单个、多个人体目标的距离轨迹跟踪;并针对联合概率数据关联算法在回波数目增多时计算量易出现爆炸现象的问题,提出一种改进的联合概率数据关联算法。该方法不仅实现了对轨迹交叉的多个人体目标的有效跟踪,而且通过与联合概率数据关联算法性能比较,二者的性能相当,但计算量却大大减少。
[Abstract]:Short-range human tracking is very important in security applications, such as airport security, detection of terrorists through walls, rescue of trapped people from debris. UWB Ultra-Wideband UWBradar has good range resolution and penetration capability. It is very suitable for short range human body tracking. On the other hand, the particularity of working environment puts forward higher requirements for UWB radar in target detection, identification and tracking. This paper focuses on the detection and tracking of human body target by UWB radar. Firstly, the system composition, signal system and signal processing flow of UWB radar are introduced, and four kinds of signals commonly used in UWB radar are analyzed in detail. Three problems, namely coupling alignment, background cancellation, and multipath effect suppression, are introduced, in which coupling alignment can improve the performance of background cancellation and suppression of coupled strong clutter. Exponential weighted background cancellation can indicate both radial and non-radial moving targets, and multipath and clutter can be separated from target echo by time window. The main purpose is to judge the existence or absence of target. When the traditional constant False Alarm CFAR method is used to detect a single human target, it is necessary to set a relatively high false alarm probability in order to keep the target information well. But at the same time, the background clutter will also increase, and in the detection of a number of close, even cross, overlapping human targets, there will be a serious target masking phenomenon. In this paper, CLEAN algorithm, which is usually used in image processing, is applied to human body target detection of UWB radar. Compared with CFAR algorithm, clear algorithm has better detection performance for multiple human body targets under different moving states. And can effectively suppress clutter, multi-path and target masking, self-masking phenomenon, and can very well retain the information of the target, Multiple human scattering points are extracted and the arrival delay of each scattering point is recorded. Based on the echo model of human body multiple scattering points, the measurement information of human body multiple scattering points is extracted by using CLEAN algorithm. Combined with nearest neighbor data association algorithm and joint probabilistic data association algorithm, the distance trajectory tracking of single or multiple human objects is realized respectively. The joint probabilistic data association algorithm is prone to explosion when the number of echoes increases. An improved joint probabilistic data association algorithm is proposed, which not only realizes the effective tracking of multiple human objects whose tracks are crossed, but also compares the performance of the joint probabilistic data association algorithm with that of the joint probabilistic data association algorithm. But the amount of calculation is greatly reduced.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN953
，

本文编号：1610004