High Spatial-spectral Resolution Image Fusion Algorithm Based on Spectral Mixture Analysis

摘要

This paper proposed a high spatial-spectral resolution image fusion algorithm for multispectral and hyperspectral images based on spectral mixture analysis. For this purpose, correspondence was established between high-spectral-resolution data and high-spatial-resolution data by extracting the same endmember. Specifically, Linear mixture model was applied to unmixing images, and the inverse process was used to do the image fusion. To test the algorithm, Hyperion and SPOT data were chose to do the fusion experiment, taking PSNR and SA as evaluation indexes. Results demonstrate the proposed algorithm a good performance on both spectral fidelity and spatial enhancement ability.