Novel and fast EMD-based image fusion via morphological filter

This paper presents a novel and fast EMD-based (empirical mode decomposition-based) image fusion approach via morphological filter. Firstly, we develop a multi-channel bidimensional EMD method based on morphological filter to conduct image fusion. It uses the morphological expansion and erosion filters to compute the upper and lower envelopes of a multi-channel image in the sifting processing, and can decompose the input source images into several intrinsic mode functions (IMFs) with different scales and a residue. It significantly improves the computation efficiency of EMD for multi-channel images while maintaining the decomposition quality. Secondly, we adopt a patch-based fusion strategy with overlapping partition to fuse the IMFs and residue instead of the pixel-based fusion way usually used in EMD-based image fusion, where an energy-based maximum selection rule is designed to fuse the IMFs, and the feature information extracted by IMFs is used as a guide to merge the residue. Such strategy can extract the salient information of the source images well and can also reduce the spatial artifacts introduced by the noisy characteristics of the pixel-wise maps. A large number of comparative experiments on the fusion of several commonly used image data sets with multi-focus and multi-modal images, show that our newly proposed fusion method can obtain much better results than the existing EMD-based image fusion approaches. Furthermore, it is very competitive with the state-of-the-art image fusion methods in visualization, objective metrics, and time performance. The code of the proposed method can be downloaded from: https://github.com/neepuhjp/MFMBEMD-ImageFusion.