Aiming at the fusing characteristics of Synthetic Aperture Radar (SAR) images and optical images, an algorithm for image fusion based on Non-subsampled Contourlet Transformation (NSCT) is proposed. The SAR images and optical source images are decomposed by NSCT transformation. To the low-frequency sub-images, the local entropy is taken as the measure parameter to carry out neighborhood fusion. To the top layer of the high-frequency sub-images, the fusion method of “choosing the biggest” for the regional standard deviation is adopted. As to the other layers of the high-frequency sub-images, the neighborhood correlation coefficient is taken as the threshold, and the neighborhood algorithm based on average gradient selection is adopted to carry out the fusion. Finally, NSCT inverse transformation is utilized to obtain the final fusion image. The experimental results show that the ideal fusion image and more detailed information can be obtained by using this method.