Enhancing Hyperspectral and Multispectral Image Fusion Using High Dimensional Model Representation

Hyperspectral imagery provides valuable information on reflective surfaces through its rich spectral content. However, it inherently suffers from low spatial resolution, as achieving high spatial detail requires a strong signal-to-noise ratio. Hyperspectral and multispectral image fusion is employed to generate images with high spectral and spatial resolution to address this limitation. In this study, Coupled Non-negative Matrix Factorization (CNMF) is taken into focus. We propose an efficient fusion method that integrates High Dimensional Model Representation (HDMR) with CNMF, which significantly outperforms the plain CNMF in terms of peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and mutual information (MI). Experiments are conducted in various hyperspectral and multispectral image dataset. It is observed that, compared to the fused images obtained from CNMF, the proposed method improves PSNR by up to 12 dB, SSIM by up to 0.70, and MI by up to 0.30.