【作者】 王晓宇；

【导师】 生雪莉；

【作者基本信息】 哈尔滨工程大学 ， 信息与通信工程， 2018， 硕士

【摘要】 水下多目标的跟踪是现代声呐研究的热点之一。近年来新兴的水下目标跟踪方法是检测前跟踪方法(TBD,Track Before Detect),这种方法直接对未经门限或经过低门限的多帧回波数据进行积累和联合处理。将该方法应用于声呐探测,能够充分的利用声呐预处理后的数据,提高声呐在低信噪比环境中对多目标的跟踪性能。本文首先研究了水声环境中多目标跟踪的基本内容,包括水下多目标跟踪原理以及关键问题,概述了水下多目标跟踪的方法以及经典算法,分析了TBD相对于跟踪前检测(DBT,Detect Before Track)的优势所在以及TBD技术难点,其中数据关联算法是多目标跟踪技术的核心内容。给出了本文仿真分析部分所用到的目标运动模型以及仿真评价指标。然后,本文研究了基于多假设跟踪的检测前跟踪算法(MHT-TBD,Multi Hypothesis Tracking-TBD)的原理,将算法原理用流程框图的形式清晰地表达出来,利用MATLAB建模仿真来对MHT-TBD算法的性能进行评估。采用改进的交互多模多假设跟踪(IMM-MHT-TBD,Interacting Multiple Model-MHT-TBD)算法,优化了算法效率并提高了实用性。最后,本文研究了基于动态规划的检测前跟踪算法(DP-TBD,Dynamic Programming-TBD)。先是对水下单目标情况做了分析,进行了建模与仿真验证,给出了相应的结论。该算法具有门限设置困难、能量损失较大以及不能在信噪比很低的环境中很好的跟踪目标等局限性。随后本文采用了改进的动态规划算法,能够解决在水下单目标跟踪过程中的如上局限性问题。为了跟踪水下多目标,对单目标DP-TBD算法做出相应改进,给出区域最大值法(RM-DP-TBD,Regional Maximum-DP-TBD),仿真结果表明,上述方法确实能够对水下多目标进行有效跟踪。更多还原

【Abstract】 The detection and tracking of underwater multi targets is one of the hotspots in modern sonar research.In recent years,the new method on underwater target detection and tracking is the tracking before detection method(TBD).This method accumulates and processes the multi frames of echo data without threshold or low threshold directly.Applying this method to sonar detection,we can make full use of the original sonar data and improve the performance of sonar in low SNR environment.This paper first studies the basic contents of multi-target tracking in underwater acoustic environment,including the principles and key issues of underwater multi-target tracking,summarizes the underwater multi-target tracking methods and classical algorithm,and analyzes the advantages of TBD compared with DBT and the difficulties of TBD technology.The problem of data association is described in detail which is called the core technology in MHT-TBD,and the target motion models and simulation evaluation index used in the simulation part are given.Then,this paper introduces the principle of Multiple Hypothesis Tracking based on Tracking Before Detection(MHT-TBD),and gives the flow chart of the algorithm,and verifies the feasibility of the algorithm through MATLAB simulation.The improved interactive multimode MHT-TBD(IMM-MHT-TBD)algorithm is used to optimize the efficiency and improve the practicability of the algorithm.Finally,this paper introduces a dynamic programming based on Tracking Before Detection(DP-TBD).First,we analyze the situations of underwater single target,and get the corresponding conclusion by modeling and simulation.The algorithm has the limitations of difficult threshold setting,large energy loss and bad performance in the environment with low SNR.In this paper,an improved dynamic programming algorithm is adopted to solve the problem of limitations in the process of underwater single target tracking.In order to carry out the detection and tracking of underwater multi targets by using the DP-TBD algorithm,the DP-TBD must be improved.The simulation results show that the maximum regional method(MR-DP-TBD)method can effectively track and detect the underwater multi targets.更多还原