Remote Image Fusion Based on PCA and Dual Tree Compactly Supported Shearlet Transform

This paper presents a novel remote sensing image fusion algorithm, which implements panchromatic sharpening of multispectral data through application of the principal component analysis (PCA) transform and the dual-tree compactly supported shearlet transform (DT CSST). Shearlet transforms provide near optimal representation of the anisotropic features of an image. The compactly supported shearlet transform (CSST) as the spatial domain implementation of discrete shearlet transform may represent the directions by the way of dilation operations directly in spatial domain. Since most of the prominent features of images, such as edges and regions, have finite regions in spatial domain, CSST is very suitable for image fusion. However, the conventional CSST, which is shift-variant, causes distortions in fused images. With the embedded structure of dual-tree (DT) in the CSST, the shift-variant properties can be effectively reduced. The evaluation results of experiments indicated that the proposed method is superior to other PCA and multi-scale transform based methods such as curvelet, band limited shearlet transform (BLST), which is an invariant shearlet transform,à trous wavelet transform, the dual-tree complex wave transform (DTCWT), and the discrete wavelet transform (DWT).