Multi-channel Satellite Cloud Image Fusion in the Shearlet Transform Domain and Its Influence on Typhoon Center Location

Abstract

A multi-channel satellite cloud image fusion method by the shearlet transform is proposed. The Laplacian pyramid algorithm is used to decompose the low frequency sub-images in the shearlet domain. It averages the values on its top layer, and takes the maximum absolute values on the other layers. In the high frequency sub-images of the shearlet domain, fusion rule is constructed by using information entropy, average gradient and standard deviation. Next, a nonlinear operation is performed to enhance the details of the fusion high frequency sub-images. The proposed image fusion algorithm is compared with five similar image fusion algorithms: the classical discrete orthogonal wavelet, curvelet, NSCT, tetrolet and shearlet. The information entropy, average gradient and standard deviation are used objectively evaluate the quality of the fused images. In order to verify the efficiency of the proposed algorithm, the fusion cloud image is used to determine the center location of eye and non-eye typhoons. The experimental results show that the fused image obtained by proposed algorithm improve the precision of determining the typhoon center. The comprehensive performance of the proposed algorithm is superior to similar image fusion algorithms.