基于辐射噪声特征的舰船目标识别分类方法综述

Review of Ship Target Recognition and Classification Methods Based on Radiated Noise Feature

摘要:

在水下声学研究中，对舰船目标进行识别和分类是一个非常重要的课题。舰船辐射噪声中蕴含着大量的特征量，可以对舰船目标进行识别和分类。首先，本文介绍四个水声目标辐射噪声数据库，这些数据库采集数据量较大，采集信息记录完整，不仅包含舰船目标，还有海洋生物、环境噪声等水下目标，可以为不同需求的研究人员提供数据支持。其次，本文总结水声领域内常用的舰船辐射噪声特征提取技术，从线谱识别、小波分析和非线性识别三个方向，对目前较为成熟的舰船辐射噪声识别技术进行总结。最后，文章从传统的识别技术和深度学习两个方向，分析了舰船辐射噪声识别方向的技术发展。传统的识别技术主要涉及到支持向量机、聚类分析等技术。深度学习是近年飞速发展的识别技术,是舰船目标识别的一个热点研究方向。本文旨在为舰船目标分类识别的研究者提供一个全面的研究现状和发展趋势的综述，为舰船目标分类识别的进一步研究和应用提供参考。

Abstract:

In underwater acoustic research, the recognition and classification of ship targets is a very important issue. Ship radiated noise contains a large number of features, which can identify and classify ship targets. Firstly, this paper introduces four underwater acoustic target radiation noise databases. These databases collect a large amount of data and record the collected information completely. They include not only ship targets, but also underwater targets such as marine organisms and environmental noise, which can provide data support for researchers with different needs. Secondly, this paper summarizes the commonly used ship radiated noise feature extraction technology in the field of underwater acoustics. From the three directions of line spectrum recognition, wavelet analysis and nonlinear recognition, the current mature ship radiated noise recognition technology is summarized. Finally, the paper analyzes the technical development of ship radiated noise recognition from the two directions of traditional recognition technology and deep learning. The traditional recognition technology mainly involves support vector machine, cluster analysis and other technologies. Deep learning is a rapidly developing recognition technology in recent years, and it is a hot research direction of ship target recognition. This paper aims to provide a comprehensive overview of research status and development trends for researchers of ship target classification and recognition.