Abstract: For the fusion problem of the multi-focus images of the same scene, a novel algorithm is proposed based on the nonsubsampled contourlet transform (NSCT). The selection principles of the low frequency subband coefficients and bandpass directional subband coefficients are discussed respectively. For choosing the low frequency subband coefficients, we present a scheme based on the local area energy of image gradient combined with the weighted average scheme, which can not only restrain the noise effectively but also select the coefficients properly. When choosing the bandpass directional subband coefficients, we make best use of such NSCT characteristics as directional sensitivity and same size between subband image and the original image, and define the directional contrast in the contourlet domain. Then we present a selection principle based on the directional contrast. The algorithm has been used to merge several sets of multi-focus images. The experimental results indicate that the proposed approach can avoid the introduction of artifacts and can significantly outperform the traditional wavelet-transform-based image fusion method in terms of both visual quality and objective evaluation criteria.