A novel image fusion method based on SGNN

Abstract

Multi-sensor image fusion is a challenging research field, which is a issue to be further investigated and studied. Self-Generating Neural Networks (SGNNs) are self-organization neural network, whose network structures and parameters need not to be set by users, and its learning process needs no iteration. An approach of image fusion using a SGNN is proposed in this paper. The approach consists of pre-processing of the images, clustering pixels using SGNN and fusing images using fussy logic algorithms. The approach has advantages of being wieldy to be used by users and having high computing efficiency, The experimental results demonstrate that the MSE (mean square error) of this approach decreases 30%-60% than those by Laplacian pyramid and discrete wavelet transform approaches. © Springer-Verlag Berlin Heidelberg 2005.