Wavelet fusion method, wavelet basis and decompose levels play an important role in the fusion algorithm of multi-focus images based on wavelet transform. A new image fusion method based on local wavelet energy is proposed, some wavelet types and levels are analyzed and compared. The fusion parameters such as entropy, peak signal noise rate and spatial frequency are used to evaluate the fusion results. The results show that the new fusion method can get a better fusion results, bior2.2 wavelet basis and 4~6 levels are the best in image fusion, which provides some recommendation for wavelet basis selection.