Review of EMD-Based Image Fusion

EMD is a new multi-scale analysis method, which is superior in non-stationary and non-linear signal analysis. It is considered as a great breakthrough of Fourier-based signal analysis. Because of its ability to separate spatial frequencies, it is natural to consider EMD for the problem of heterogeneous image fusion. The intrinsic decomposition nature can adaptively decompose images into a series of modes representing different local features of images. Fusing modes with different features is helpful for image interpretation and analysis. This paper introduces the principle of EMD and its extensions, including BEMD, CEMD, and EEMD, as well as their application in image fusion. It also makes a discussion on the advantages, key issues and difficulties of EMD when applied to image fusion.