Bivariate EMD-based image fusion

The empirical mode decomposition (EMD) algorithm is a fully data-driven method which is used to perform an adaptive decomposition of nonlinear and nonstationary signals. It has been recently illustrated that its complex extensions can be used to carry out fusion of multiple images. This is possible because the complex EMD allows comparison between common frequency scales, by aligning them within a single complex IMF. In this paper, complex extensions of EMD are proposed for the fusion of two images; the fusion methodologies are presented for both gray-level and RGB based color images. The potential of the proposed scheme is highlighted by showing its superiority to wavelet based fusion schemes, through simulations on real world multi-exposure images.