Hyperspectral image multiresolution fusion based on local information entropy

Hyperspectral image with high spectral resolution provides more information than multispectral image. But its large data volumes bring many difficulties for effective information extraction. In recent years, information fusion technique has been paid great attention and used widely in the area Of remote sensing image processing. Taking into consideration of hyperspectral image characteristics, a new multiresolution fusion method based on local information entropy (LIE) is proposed in this paper. The key technique is based on the fact that the larger the entropy is, the more information the image contains. The Relative local information entropy (RLIE) is used for determining the weights of hyperspectral images in fusion. The experimental results on AVIRIS data show that the new method, using a few feature images but containing most of information in subspaces, is suitable for the hyperspectral image fusion display and classification.