基于提升小波变换的MFCC在目标识别中的应用  

Application of the lifting wavelet transform based MFCC in target identification

作　　者：石超雄 李钢虎[1] 何会会[1] 赵妮[2] 

机构地区：[1]西北工业大学航海学院,陕西西安710072 [2]西安邮电大学通信与信息工程学院,陕西西安710072

出　　处：《声学技术》2014年第4期372-375,共4页Technical Acoustics

摘　　要：水下环境噪声对被动声呐目标的分类识别影响显著,为了提高水声信号识别系统对环境噪声的鲁棒性,采用提升小波变换方法提取MFCC特征,对水下被动声呐目标进行分类和识别。仿真实验表明,与传统MFCC和小波变换的MFCC相比,在信噪比相近的情况下,提升小波变换方法提取MFCC具有识别率较高、对噪声鲁棒性较好的显著优点。The underwater noise could badly affect the results of passive sonar classification and target recognition. In order to improve the robustness of underwater acoustic signal identification system against environmental noise, a new MFCC feature extraction method based on lifting wavelet transform is proposed. Experimental results show that, compared with the traditional MFCC, the wavelet transform MFCC method for underwater acoustic signal classification and identification works better under the same SNR, which significantly improves the robustness against noise.

关 键 词：提升小波变换 MEL频率倒谱系数 分类识别 

分 类 号：TB556[理学—物理]