Two-Step Sparse Coding for the Pan-Sharpening of Remote Sensing Images

Remote sensing image pan-sharpening is an important way of enhancing the spatial resolution of a multispectral (MS) image by fusing it with a registered panchromatic (PAN) image. The traditional pan-sharpening methods often suffer from color distortion and are still far from being able to synthesize a real high-resolution MS image, as could be directly acquired by a better sensor. Inspired by the rapid development of sparse representation theory, we propose a two-step sparse coding method with patch normalization (PN-TSSC) for image pan-sharpening. Traditional one-step sparse coding has difficulty in choosing dictionary atoms when the structural information is weak or lost. By exploiting the local similarity between the MS and PAN images, the proposed sparse coding method deals with the dictionary atoms in two steps, which has been found to be an effective way of overcoming this problem. The experimental results with IKONOS, QuickBird, and WorldView-2 data suggest that the proposed method can effectively improve the spatial resolution of a MS image, with little color distortion. The pan-sharpened high-resolution MS image outperforms those images fused by other traditional and state-of-the-art methods, both quantitatively and perceptually.