The work aims to solve the defect such as poor fusion image visual effect and low robustness induced by mostly achieving the image fusion on the pixel gray space of current image fusion algorithm. A new multi-focus image fusion algorithm based on improved shearlet transform coupling frequency characteristic was proposed in this paper. Firstly, the shearlet transform (ST) and the non-subsample wavelet transform ((NSWT)) were fused to form the improved shearlet transform (ST-NSWT), and the ST-NSWT transform was used to decompose the source image to obtain the low and high frequency subband coefficients of the image. Then the regional energy model was constructed to measure the dependency between the low frequency subband coefficients of the source images, and to complete the fusion of low frequency subbands. Finally, through analysis on the frequency characteristics of high frequency subbands, the variance model, the average gradient model and the spatial frequency model were established to measure the gray correlation, resolution correlation and activeness correlation of the source image, so as to complete the fusion of high frequency subbands, and finally the fuse image was accomplished by inverse ST-NSWT transform. Compared with the current multi-focus image fusion algorithm, the proposed algorithm could preserve more details and edge information in a better way, which makes the fusion image have better visual effect. The proposed algorithm has better fuse quality and it can be used in such fields as remote sensing and packaging & printing detection.