Adaptive Fusion of Remote Sensing Images with Sparse Representation

In this paper, a novel sparse representation-based image fusion method is proposed. This method uses the base vectors corresponding to the nonzero components of sparse coefficients as image features. Firstly, the common and respective base vectors are separated. Then the corresponding sparse coefficients are consequently weighted. Finally, the fused image is reconstructed from the combined sparse coefficients and the overcomplete dictionary. Our method fuses the common and respective features separately, so it can overcome the problem of the loss of clarity in the fused image. Furthermore, since sparse representation has been significantly successful in the development of image denoising algorithms, our method can carry out image denoising and fusion simultaneously. Compared with four state-of-the-art methods, the performance of the proposed method is better.