基于改进的MFCC战场被动声目标识别

Battlefield Passive Acoustic Target Identification Based on Improved MFCC

从战场声信号与语音信号特征的相似性出发,提出基于MFCC参数的战场声目标识别方法。针对战场环境存在强噪声干扰情况,提出一种改进的MFCC特征参数(DWTMFCC)提取方法,该方法将小波分析和Mel倒谱分析结合,提高了特征参数的鲁棒性。仿真结果表明:在噪声条件下,利用DWTMFCC参数进行声目标识别,平均识别率比MFCC参数高出3.134个百分点,信噪比为5dB时,识别率仍达到93.67%。

Abstract：Beginning with the similarity between battlefield acoustic signal and speech signal, a new battlefield acoustic target identification method was proposed based on Mel frequency cepstrum coefficient（MFCC）. As stronger noise exists in battlefield, an improved MFCC parameter （DWTMFCC） was presented. The DWTMFCC parameter extraction combined discrete wavelet transform （DVCT） analysis with Mel frequency cepstrum analysis, and its robustness was improved. The simulation result shows that the identification ratio using DWTMFCC is improved by 3. 134% compared with MFCC under noisy environment. When the SNR is 5dB, the identification ratio is up to 93.67%.