An improved-SFIM fusion method based on the calibration process

The fusion of multispectral/hyperspectral image (MSI/HSI) and panchromatic image (PI) is a crucial and useful issue. Among the different fusion methods, smoothing-filter-based intensity modulation (SFIM) fusion technique is a simple yet effective model. It is proposed based on a simplified solar radiation and land surface reflection model. However, the calibration process, which is important in the remote sensing, is neglected. Meanwhile, in the fusion model, instead of using the accurate radiance value, the digital number (DN) value of optical image is improperly used. It caused the distortion of color in the fused images. Therefore, in the letter, we propose an improved-SFIM (ISFIM) by exploiting the calibration process. In the model, the range of DN values of the fused data are properly constrained in the calibration process. Experiments on the MSI and HSI are presented along with the fusion results obtained by discrete wavelet transform, Gram–Schmidt transform, original SFIM techniques and the proposed ISFIM. The results illustrate that, the ISFIM behaves better both in visual analysis and objective indices than existing methods.