机动扩展目标联合跟踪与分类算法研究

发布时间：2018-10-05 14:06

【摘要】：随着雷达、红外等传感器分辨率的不断提高,扩展目标跟踪受到了国内外学者的广泛关注。近年来,由于计算简单且能够跟踪扩展目标,基于随机矩阵的扩展目标联合跟踪与分类算法研究受到关注。本论文以贝叶斯滤波和随机矩阵为理论基础,针对椭圆和非椭圆两类扩展目标,研究机动扩展目标联合跟踪与分类方法,具有重要的理论意义和应用价值。论文主要研究内容如下:1.研究了机动扩展目标跟踪的基础理论,包括目标运动模型、贝叶斯滤波理论、随机矩阵算法和多模型算法等。以常规运动模型为基础,研究了机动扩展目标跟踪的多模型算法;以贝叶斯滤波为理论框架,研究了基于随机矩阵的目标扩展状态估计和目标跟踪算法;为了同时估计机动扩展目标的运动状态和扩展状态,研究了基于随机矩阵的机动椭圆扩展目标跟踪算法。2.针对机动椭圆扩展目标,结合交互多模型,提出一种基于随机矩阵的机动椭圆扩展目标联合跟踪与分类算法。该算法以贝叶斯滤波为理论框架,结合目标的大小、形状、方向等先验信息,利用随机矩阵实时估计目标的状态,同时采用多模型算法处理椭圆扩展目标的机动情况。仿真实验表明,所提算法可实时估计目标的运动状态、扩展状态和类状态等信息,实现机动椭圆扩展目标的联合跟踪与分类。3.针对机动非椭圆扩展目标,结合交互多模型,提出一种基于随机矩阵的机动非椭圆扩展目标联合跟踪与分类算法。该算法将非椭圆目标近似为多个相互独立的子椭圆,每个子椭圆由随机矩阵表示,其中与目标有相同方向的子椭圆作为主椭圆,并用结构信息表示子椭圆与主椭圆之间的关系;采用已提出的基于随机矩阵的机动椭圆扩展目标联合跟踪与分类算法估计主椭圆的状态;利用结构信息估计各子椭圆的状态,实现机动非椭圆扩展目标的联合跟踪与分类。仿真实验表明,所提算法可改善椭圆近似目标带来的信息丢失问题,在有效跟踪非椭圆扩展目标的同时,能实时估计目标的运动状态、扩展状态以及类状态等信息。
[Abstract]:With the increasing resolution of radar, infrared sensors and other sensors, extended target tracking has been widely concerned by scholars at home and abroad. In recent years, due to the simple computation and the ability to track extended targets, the research of joint tracking and classification algorithm based on stochastic matrix has attracted much attention. Based on Bayesian filtering and stochastic matrix, this paper studies the joint tracking and classification method of maneuvering extended targets for elliptic and non-elliptic extended targets, which has important theoretical significance and application value. The main contents of this paper are as follows: 1: 1. The basic theories of maneuvering extended target tracking are studied, including target motion model, Bayesian filtering theory, stochastic matrix algorithm and multi-model algorithm. Based on the conventional motion model, the multi-model algorithm of maneuvering extended target tracking is studied, and the extended state estimation and target tracking algorithm based on random matrix are studied based on Bayesian filter. In order to simultaneously estimate the moving state and extended state of maneuvering extended target, a maneuvering elliptic extended target tracking algorithm based on random matrix is studied. For maneuvering elliptic extended targets, an algorithm for joint tracking and classification of maneuvering elliptic extended targets based on random matrix is proposed. The algorithm takes Bayesian filtering as the theoretical framework and combines the prior information such as the size, shape and direction of the target. The random matrix is used to estimate the state of the target in real time. At the same time, the multi-model algorithm is used to deal with the maneuver of the elliptical extended target. The simulation results show that the proposed algorithm can estimate the moving state, extended state and class state of the target in real time, and realize the joint tracking and classification of maneuvering elliptical extended targets. A joint tracking and classification algorithm of maneuvering non-elliptic extended targets based on random matrix is proposed for maneuvering non-elliptic extended targets. The algorithm approximates the non-elliptic object as several independent sub-ellipses, each subellipse being represented by a random matrix, in which the sub-ellipse with the same direction as the target is taken as the main ellipse, and the relation between the sub-ellipse and the principal ellipse is expressed by structural information. The proposed joint tracking and classification algorithm for maneuvering elliptic extended targets based on random matrix is used to estimate the state of the main ellipse, and the structure information is used to estimate the states of each sub-ellipse to realize the joint tracking and classification of maneuvering non-elliptical extended targets. Simulation results show that the proposed algorithm can improve the problem of information loss caused by elliptical approximate targets. The proposed algorithm can effectively track non-elliptical extended targets and estimate the moving state, extended state and class state of the target in real time.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN713

【相似文献】