基于一维距离像的舰船目标分类研究

发布时间：2018-09-18 18:11

【摘要】：基于一维距离像或高分辨距离像(High Range Resolution Profile,HRRP)的舰船目标识别技术是雷达数据处理领域的一个重要研究方向。海上舰船作为一种移动目标,其一维距离像所具有的平移敏感性、姿态敏感性以及幅度敏感性问题是该领域的技术难点。本论文针对如何提高基于HRRP的舰船目标识别率以及如何搭建一个实用型的舰船目标识别平台来展开研究,主要工作如下:(1)介绍了该领域的研究背景和意义,并对一维距离像的特征提取、多特征融合以及分类器设计的国内外研究现状作了简单介绍。(2)研究了一维距离像的数据预处理技术。首先,对原始HRRP数据进行低通滤波以降低海杂波背景的噪声干扰,提高信噪比。然后,对滤波后的HRRP数据进行归一化处理,减少了HRRP的幅度敏感性。最后,通过估计船头和船尾的位置来提取目标区域,解决了HRRP的平移敏感性问题。(3)针对HRRP的姿态敏感性问题,本论文根据传统的特征提取方法,提取多个具有姿态不敏感的特征来进行舰船目标识别。为了提高识别率,本论文提出一种基于紧致性测度加权的舰船类型识别算法,通过该算法可反映出不同特征在识别过程中的权重差异。另外,本论文对传统的基于Fisher准则的特征选择方法进行改进,提出了一种基于特征互补性的特征选择方法,从所提取的特征集中选择出一个最优特征子集,并通过实验证明了该算法可以使用更少的特征得到更高的识别率。(4)为尽可能的模拟真实场景下的舰船目标识别过程,本论文设计了一个集多功能于一体的识别平台。该平台实现的主要功能有:航迹以及HRRP数据的接收或发送、数据的实时显示、成像数据的处理、数据库存储查询、系统设置等功能。
[Abstract]:Ship target recognition based on one dimensional range profile or high resolution range profile (High Range Resolution Profile,HRRP) is an important research direction in the field of radar data processing. As a moving target, the translation sensitivity, attitude sensitivity and amplitude sensitivity of one dimensional range profile are the technical difficulties in this field. This paper focuses on how to improve the rate of ship target recognition based on HRRP and how to build a practical platform for ship target recognition. The main work is as follows: (1) the research background and significance of this field are introduced. The feature extraction, multi-feature fusion and classifier design of one-dimensional range profile are briefly introduced. (2) the data preprocessing technology of one-dimensional range profile is studied. Firstly, the original HRRP data are filtered by low pass filter to reduce the noise interference of sea clutter background and improve the signal-to-noise ratio (SNR). Then, the filtered HRRP data is normalized to reduce the amplitude sensitivity of HRRP. Finally, the translation sensitivity of HRRP is solved by estimating the position of bow and stern. (3) aiming at the attitude sensitivity of HRRP, the traditional feature extraction method is used in this paper. A number of pose insensitive features are extracted for ship target recognition. In order to improve the recognition rate, this paper proposes a warship type recognition algorithm based on compactness measure weighting, which can reflect the weight difference of different features in the recognition process. In addition, this paper improves the traditional feature selection method based on Fisher criterion, proposes a feature selection method based on feature complementarity, and selects an optimal feature subset from the extracted feature set. Experiments show that the algorithm can use fewer features to achieve a higher recognition rate. (4) in order to simulate the process of ship target recognition in real scene as much as possible, this paper designs a multi-functional recognition platform. The main functions of the platform are as follows: receiving or sending the track and HRRP data, displaying the data in real time, processing the imaging data, storing and querying the database, setting up the system, etc.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：E925;TN957.52

【参考文献】