To preserve defined edges of the fused multi-focus image while enriching the detail features of the image, a novel algorithm based on the fast finite shearlet transform (FFST) and the guided filter is proposed. Firstly, the original images are decomposed into low frequency subband coefficients and bandpass direction subband coefficients by using FFST. Then, in the fusion of the low frequency coefficient, a novel Sum-Modified-Laplacian (NSML) is defined, and a selection scheme of low frequency coefficients is designed based on regional NSML. Due to the rich detail information of high frequency coefficient, we present a regional weighting energy fusion algorithm based on the guided filter. Finally, the final fused image is produced by inverse FFST. Comparison experiments are performed on different image sets, and experimental results demonstrate that the proposed algorithm performs better in both subjective and objective qualities.