摘要
利用分数傅里叶变换(FrFT)在时频域对信号的混合表示的特点,将离散分数傅里叶变换(DFrFT)和相关向量机(RVM)应用于毫米波(MMV)高分辨雷达一维距离像识别。使用Fisher准则确定DFrFT的阶数α,将一维距离像进行α阶DFrFT变换,获得信号的特征量,然后利用RVM网络进行分类识别。实验结果表明,该方法是一种可行有效的特征选择方法,具有较高的识别率。
Since fractional Fourier transform(FrFT)is a mixed expression of signal in time-frequency domain,the discrete fractional Fourier transform(DFrFT)and relevance vector machine were introduced into one-dimension range profile identification of high-resolution millimeter waves(MMW)radar.The order α of discrete fractional Fourier transform was confirmed by Fisher criterion.The features of range profiles were extracted by discrete fractional Fourier transform at order α and classified by the relevance vector machine(RVM)network.The experimental results show that the method is feasible and has higher recognition probability.
出处
《兵工学报》
EI
CAS
CSCD
北大核心
2010年第7期902-905,共4页
Acta Armamentarii
关键词
信息处理技术
离散分数傅里叶变换
FISHER准则
相关向量机
特征提取
information processing
discrete fractional Fourier transform
Fisher criterion
relevance vector machine
feature extraction