Region-based fusion of infrared and visible images using Bidimensional Empirical Mode Decomposition

Region-based image fusion schemes have been studied a lot, but they are all based on some common decomposition, such as pyramid, wavelet and contourlet transform. In this paper, we present a novel region-based image fusion scheme using BEMD (Bidimensional Empirical Mode Decomposition) for infrared and visible images. BEMD is a new 2D signal analysis method extended from EMD and it decomposes the signal into a series of IMFs (Intrinsic Mode Functions) from finest to coarsest. Region segmentation is of vital importance in the fusion process. Real images are always intensity inhomogeneous, e.g. infrared and visible images, so we use an LBF (Local Binary Fitting) model which aims at segmenting intensity inhomogeneous images to extract our regions. Experiments show that the proposed fusion scheme works effectively compared with traditional fusion schemes.