SAR and Multispectral Image Fusion Algorithm Based on Sparse Representation and NSST

摘要

Aiming at the problem of spectral distortion and texture detail loss in synthetic aperture radar (SAR) image and multi-spectral (MS) image fusion, an image fusion algorithm combining sparse representation (SR) and non-subsampled Shearlet transform (NSST) is proposed. The algorithm uses the multi-scale, multi-directional and translation-invariant characteristics of NSST to transform and decompose the luminance components of SAR images and multi-spectral images. Then, the low-frequency sub-band is represented by SR, and the fusion is performed by an energy-adaptive method. The high-frequency sub-band is fused with the correlation coefficient as the saliency index, and finally the fused image is obtained by inverse transformation. The simulation experiments show that the proposed algorithm effectively preserves the subject information and feature information of the source image, so that the contrast of the fused image is significantly improved, the image outline is clear, and the overall sharpness. The spectral resolution and spatial resolution are closer to the fused reference image.