Kernel ICA Feature Extraction for Anomaly Detection in Hyperspectral Imagery  

作　　者：ZHAO Chunhui WANG Yulei MEI Feng 

机构地区：[1]College of Information and Communication Engineering, Harbin Engineering University, Harbin 150001, China

出　　处：《Chinese Journal of Electronics》2012年第2期265-269,共5页电子学报（英文版）

基　　金：Manuscript Received Nov. 2010; Accepted Oct. 2011. This work is supported by the National Natural Science Foundation of China （No.61077079, No.60802059）, the Ph.D. Programs Foundation of Ministry of Education of China （No.20102304110013） and the Excellent Academic Leader Foundation of Harbin City in China （No.2009RFXXG034）.

摘　　要：A kernel-based independent component analysis algorithm, which combines Kernel principal com- ponent analysis （KPCA） and Independent component analysis （ICA） is proposed for anomaly detection in hyper- spectral imagery. Firstly, KPCA is performed on a feature space associated with the original hyperspectral data space via a certain nonlinear mapping function to whiten data and fully mine the nonlinear information between spec- tral bands. Then~ ICA seeks the projection directions in the KPCA whitened space for making the distribution of the projected data mutually independent. Finally, RX de- tector is performed on the projected data to locate the anomaly targets. The kernel ICA algorithm saved the nonlinear information on dimension reduction in hyper- spectral data and made the extracted features mutually independent, so improved the performance of RX detector in hyperspectral data. Numerical experiments are con- ducted on real hyperspectral images. Using receiver oper- ating characteristic curves, the results show the improved performance and reduction in the false-alarm rate.

关 键 词：Anomaly detection Hyperspectral im-agery Independent component analysis （ICA） Kernel-based method Principal component analysis （PCA） ReedXiaoli （RX）. 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置] TN911.7[自动化与计算机技术—控制科学与工程]