Spatial fusion algorithm of multispectral and high-resolution panchromatic images

Abstract

Image merging based on wavelet transformation requires a large computation complexity and may cause a severe lost of the spectral information of the multispectral image. A new spatial image fusion algorithm based on fuzzy theory is proposed in this paper. Through calculating the neighborhood homogeneous measurement (NHM) with the fuzzy entropy, the structure information of the high-resolution panchromatic image was obtained. Two thresholds were defined for different rules of fusion. Therefore, different fusion operations were taken accordingly. Experimental results demonstrate that the proposed algorithm enhances spatial detail information of the merged image greatly and has a better ability to preserve spectral information of the original multispectral image, compared with a wavelet image merging algorithm.