摘要
特征提取是目标识别的关键技术,本文提出了一种基于时频分布重排的Wigner-Ville分布(WD)特征提取方法。选择WD作为研究问题的切入点,使用核函数方法,通过平滑伪Wigner-Ville分布(SPWD)抑制交叉项;为解决引入核函数导致信号时频聚集性下降问题,我们采用时频重排方法来增强目标的时频聚集性,最后用海上实验数据对本文方法进行了有效性检验。实验结果表明,本文提出的SPWD时频重排特征提取方法具有:(1)可有效抑制WD固有的交叉项;(2)增强目标回声信号时频聚集性;(3)可明显地提高水下目标识别率。该方法具有一定的应用推广价值,为解决水下目标识别问题提供了很好的思路。
Feature extraction is a key technology for target classification. In this paper a Wigner-Ville distribution feature extraction method based on time-frequency reassignment theory was proposed. The kernel function method and the smoothed pseudo Wigner-Ville distribution（SPWD） were used to restrain the cross items. To solve the problem of which the kernel function method may cause signal time-frequency aggregation decrea- sing, a time-frequency distribution reassignment method was applied to enhance the target time-frequency ag- gregation. Finally, the effectiveness of the method was testified by sea experiment data. The experimental results show that the advantages of SPWD time-frequency distribution reassignment feature extraction method are as follows ： （ 1 ） restraining the cross items effectively ; （2） enhancing target echo time-frequency aggregation ; and （3） increasing target classification rate. This paper provides a good method for underwater target classification which has promising application prospects.
出处
《应用声学》
CSCD
北大核心
2014年第1期87-94,共8页
Journal of Applied Acoustics
基金
国家重点实验室基金项目(51303020307-8
4010201030301)
