Research on Performance Evaluation of SAR and Optical Image Fusion Methods Based on GF-3 Satellite Images

The imaging mechanisms of SAR sensors and optical ones are different. Accordingly, there are differences between features of SAR and optical images. The fusion methods for SAR satellite images and optical ones mainly include wavelet transform, HIS transform, PCA transform, and so on. Each transformation method has different performances on their fused images. In the paper, the performance evaluation method of SAR and optical image fusion methods based on real-world satellite images is studied. Firstly, SAR and optical image fusion process is introduced, including image preprocessing, image registration and image fusion. Secondly, various evaluation indicator, such as average gradient (AG), distortion, entropy, mutual information (MI), universal image quality index (UIQI) and edge mutual information (EMI) are provided. Finally, taking Yangzhong Island region in China as an experimental site, multiple image fusion methods were used to fuse a GF-3 satellite SAR image and an optical one. Evaluation criteria are provided for the applicability of image fusion methods.