Fusion of hyperspectral images based on feature images extraction and contourlet analysis

Because of the high data dimensionality of hyperspectral data, it is somehow difficult to directly apply hyperspectral images in classification and target detection. A fusion method of hyperspectral images based on feature images extraction and contourlet analysis is proposed. The algorithm firstly extracts feature images using subspace partition and principal components analysis (PCA), then these feature images are fused using adaptive low-high frequency complementary fusion algorithm based on contourlet transform. The experimental results show that the proposed algorithm has a high computation efficient. It could both compress hyperspectral images and well preserve the objects and background information of original scene, moreover, it outperform the traditional hyperspectral images fusion method in the spatial resolution improvement.