The objective of image fusion is to combine the source images of the same scene to form one composite image that contains a more accurate description of the scene than any one of the individual source images. A multiresolution image fusion method and its modified method using the local gradient of low frequency coefficients are proposed in this paper. To fuse images using the wavelet transform, the new methods are implemented to select low frequency coefficients, compared with conventional approaches. The fused images are evaluated by parameters such as the average error, the peak signal noise rate, and the root mean square error. The experimental results show that the parameters obtained with the new methods are better than those with the other conventional methods, while the optimal wavelet decomposition level of the new methods is less than that of conventional approaches.