Noise resilient image fusion based on orthogonal matching pursuit

A novel image fusion algorithm based on orthogonal matching pursuit (OMP) is proposed, which is named IFOMP algorithm and can be used for noise free or noise images. The matching pursuit algorithm provides an efficient image representation using an overcomplete dictionary. Firstly, the source images are decomposed into sub-images at different scales through Laplacian pyramid transform. The OMP method is exploited for capturing and fusing edges and texture features in the high frequency domain. Experimental results show that the proposed algorithm improves performance compared to traditional wavelet fusion with slightly increased computational complexity, and the fusion image has good robustness.