A NSP-based dictionary learning for image fusion

For the multimodal image fusion, this paper proposes a method of NSP-based and dictionary learning on image fusion. Through Non-subsampled Pyramid(NSP) filter of the image of high and low frequency decomposition, then the high frequency region of sparse representation. Clustering of parts in different images by structural similarity, which mainly composed of over-complete dictionary. Finally, Sparse coefficients are estimated by using simultaneous orthogonal matching tracking algorithm. This method is used on common dictionary learned to represent multimodal images. The experimental results show that the algorithm in the subjective and objective evaluation are raised to some extent.