Wavelet-based fusion classification for hyperspectral images

Wavelet transform and data fusion technique are widely used in many fields recently. Many special properties of them show favorable potential in the classification investigation of hyperspectral images. In this paper, a new wavelet-based fusion classification method with different weights is proposed for hyperspectral image. This new method consists of two key techniques: local feature extraction and weights determination. In order to testify the effectiveness of the proposed method, computer simulations are conducted on AVIRIS data. By comparing the classification accuracy between the wavelet-based fusion method and the classical PCA as well as current SPCT method, the new wavelet-based method is shown to provide excellent classification results 96.23% and, in every case, outperforms the other methods.