Multi-Modal Image Fusion via a Novel Multi-scale Edge-preserving Decomposition

For the traditional MSD-based infrared and visible image fusion methods, the fused images often have low contrast, and the textural details are not well preserved. This paper proposed a novel multi-scale edge-preserving decomposition method based on the guided and Gaussian filters. The small scale texture detail information and the large-scale edge information, which represented the visible feature component and the infrared feature component, respectively, can be extracted by the proposed method. In order to effectively inject the infrared information into the visible image, the large-scale edge layer is used to construct the fused weights. Experimental results show that the proposed method can not only highlight the infrared object, but also preserve the textural details as much as possible, which is superior to the existing MSD-based fusion methods both in the subjective evaluation and objective assessment. The proposed fusion method is also applicable to medical image fusion and has obtained state-of-the-art performance.