In order to overcome the bad influence caused by complex background in hyperspec tral image anomaly detection,a new anomaly detection approach is proposed.Hyperspectral data is forstly decomposed into high frequency images and low frequency image by wavelet decomposition,and the background information in hyperspectral d ate is also processed by principal component analysis (PCA).Then data after proc essing is gotten by fusing high frequency images and data after PCA.At lastly,Kerner-Reed-Xiaoli ( KRX) algorithm is used to detect the data after processing.The simulation result s show that the approach is better than other algorithms by compa ring the receiver operating characteristic (ROC) curves.