摘要
利用主成分分析理论对回声定位声波特征空间进行优化,通过小波包理论对回声定位声波进行预处理,提取能量参数,采用主成分分析法优化特征空间,减少冗余参数.以优化后的特征空间作为识别向量,通过神经网络对回声定位声波进行识别.结果表明,利用主成分分析优化特征空间,能在较高的识别率下有效降低特征空间维数.
Using the principal component analysis theory,the echolocation acoustic characteristic space was optimized.Echolocation acoustic waves were pretreated to extract energy parameters by the wavelet packet theory.The principal component analysis was adopted to optimize character spaces in order to reduce redundancy parameters.Taking optimized feature spaces as recognition vectors,echolocation acoustic waves were identified by neural network.The results showed that the principal component analysis was used to optimize feature spaces which could effectively reduce the number of dimensions of the feature space in high recognition rate.
出处
《华北水利水电学院学报》
2011年第2期79-82,共4页
North China Institute of Water Conservancy and Hydroelectric Power
基金
洛阳理工学院青年基金项目(2009QZ19)
关键词
蝙蝠
小波包
主成分分析
神经网络
识别
bat
wavelet packet
principal component analysis
neural network
recognition
