小波包分形与支持向量机在水声目标识别中的应用研究

A method based on wavelet fractal and support vector machine for underwater target recognition

水声目标分类识别是公认的水声信号处理难题,船舶辐射噪声是一种非线性非平稳信号,具有一定的混沌特性,更好地认识船舶辐射噪声的非线性性质,有助于更好地寻找有效的水声目标检测及识别算法。为了解决水声目标的分类识别问题,提出了利用小波包分形和支持向量机组合进行水声目标识别。利用小波包分解得到目标辐射噪声不同频带内信号分形维数作为特征矢量,并输入到支持向量机实现目标分类,实验结果表明,小波包分形和支持向量机的结合有比较好的分类识别效果,有一定的实际应用价值。

Underwater target recognition is one of the important and difficult topics of underwater acoustic signal processing. It is of important to extract and analyze the nonlinear and chaotic features of ship radiated noise and to recognize underwater target. To solve underwater target recognition problem, a method based on wavelet fractals and SVM（Support Vector Machine） for underwater target recognition is studied. Wavelet fractals are used to extract feature of ship radiated noise. The fractal box-counting dimensions of ship radiated noise are taken as a new parameter, which could reflect their frequency composition of the noise. And SVM is used for multi-class recognition. The experiment result shows that this method has good recognition rate.