期刊文献+

一种基于船舶辐射噪声信号改进Mel倒谱系数的目标识别方法 被引量:4

An Improved Mel-cepstrum Coefficients for Target Recognition Based on Ship Radiated Noise Signal
原文传递
导出
摘要 基于船舶辐射噪声信号Mel频率倒谱系数(MFCC)的目标类型识别是目前研究的一个热点。现有方法虽然在无噪声环境下具有较好的识别效果,但是在信噪比较低时其识别效果较差。基于此,文章提出了一种改进的提取MFCC特征参数的船舶目标识别方法,该方法在船舶辐射噪声信号的预处理阶段采用多正弦窗来代替传统使用的Hamming窗进行多窗频谱估计,经过计算得到改进的MFCC参数。试验结果表明,相比传统方法提取的MFCC参数,使用该方法提取的MFCC参数分别在不同信噪比的高斯白噪声干扰下,在BP神经网络分类器中的识别率更高,抗噪声的鲁棒性和稳定性更好。 The target identification based on MFCC(Mel Frequency Cepstrum Coefficients) of ship radiated noise signal is a hot issue. Although the present methods in noise-free environment with a good recognition results, it has a bad recognition effect at lower SNR. Based on this, in the paper an improved MFCC feature parameters for target recognition of ship is proposed. In the preprocessing stage to ship radiated noises, the multiple sinusoidal tapers are used to replace the Hamming window to predict the spectrum estimation in multi window. Then, the parameter of the improved MFCC can be achieved. The experimental results show that in different SNR of Gaussian white noise it can be found that the improved method has a better recognition effect than the traditional one. It also has a better recognition rate in BP neural network classifier and a better performance on anti-noise robustness and stability.
出处 《船舶工程》 CSCD 北大核心 2017年第1期91-95,共5页 Ship Engineering
基金 国家自然基金项目(11574120) NSFC通用技术基础研究联合基金(U1636117)
关键词 船舶辐射噪声 特征提取 MEL频率倒谱系数 多窗频谱估计 BP神经网络分类器 ship radiated noise feature extraction Mel Frequency Cepstrum Coefficients multiple sinusoidal tapers BP neural network classifier
  • 相关文献

参考文献5

二级参考文献49

共引文献66

同被引文献34

引证文献4

二级引证文献14

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部