ABSTRACT
Empirical mode decomposition (EMD), originally developed by Huang et al. (1998), is a powerful tool for decomposing nonlinear and nonstationary signals. Since this decomposition technique has been extended to analyze two-dimensional data by Song et al. (2001), bidimensional empirical mode de-composition(BEMD) developed rapidly (Han 2002, Nunes 2003, Liu 2005, among others). Some well-known methods are: BEMD of applying 1-D EMD directly along rows and then along columns of image (Han 2002), BEMD of considering direction of image texture, that is DEMD (Liu 2005), BEMD based on radial basis function interpolation (Nunes 2003); BEMD based on Delaunay triangulation and on piecewise cubic polynomial interpolation (Damerval 2005), BEMD based on finite-element technology (Xu 2006), neighborhood limited BEMD (Xu 2006), assisted signal BEMD (Xu 2011), fast and adaptive BEMD, that is FABEMD (Sharif 2008), BEMD based on PDE (Niang 2010), and so on. Among them, FABEMD is accepted as state of the art in both decomposition rate and eect, and has been used for image fusion (Ahmed 2010), image enhancement (Yang 2011, Zhang 2012), image registration and motion estimation (Mahraz 2012, Ri 2013) etc.
