Image fusion algorithm based on arbitrary shaped regions combination

The researches on images fusion have attracted more attention due to the increasing demands. A number of pixel-based and region-based image fusion algorithms are described in certain criterions. In this work, a novel region-based image fusion method is proposed, which explores pulse couples neural network (PCNN) to produce a region map. Characteristics of each region are calculated and the region-based approach is used to fuse the images in the stationary-wavelet domain. Experimental results show that the proposed algorithm outperforms other methods in objective evaluation criteria.