Second fusion algorithm of satellite images by à trous-Curvelet transform

For the multi-spectral and panchromatic image on the same scene, the second fusion algorithm by a trous-curvelet transform is constructed. Above all, the first fusion image can be acquired by using the panchromatic image to substitute the intensity component of the multi-spectral image. Then the non-negative matrix factorization (NMF) is carried out between the first fusion image and the original multi-spectral image which followed by a trous-curvelet transform on the first feature component of each image based on the theory of the maximum variance. For decomposition coefficients, the new algorithm based on weighted regional similarity is introduced. Finally, the ultimate fusion image can be acquired through inverse NMF on the new first feature component. Experimental results show that the new method can get better effect on maintaining spectral information and enhancing space quality than other methods such as wavelet transform.