Higher Order Feature Set For Underwater Noise Classification

Signal Processing: An International Journal

Mohankumar K¹  P.R. Saseendran Pillai¹  Supriya M.H¹ 

[1] $$

关键词: Bispectrum;    Bicoherence;    SVM;    HOS;    Target Classification.;   

DOI  :  

学科分类：物理（综合）

来源: Computer Science Journals

【 摘 要 】

The development of intelligent systems for classification of underwater noise sources has been a field of research interest for decades. Such systems include the extraction of features from the received signals, followed by the application of suitable classification algorithms. Most of the existing feature extraction methods rely on the classical power spectral methods, which may fail to provide information pertaining to the deviations from linearity and Gaussianity of stochastic processes. Hence, many recent research efforts focus on higher order spectral methods in order to prevail over such limitations. This paper makes use of bispectrum, which is a higher order spectrum of order three, in order to extract a set of robust features for the classification of underwater noise sources. An SVM classifier is used for evaluating the performance of the feature set.

【 授权许可】

Unknown