An image fusion algorithm just using multiscale edges of original imag es was proposed. There are three steps in this algorithm. Firstly, multiscale ed g es of the original images were detected. Secondly, these multiscale edges were fus ed to get the fusion image′s multiscale edges according to the fusion rule, max i mizing edge correlation between the fusion image and original images. At last, t he fusion image was reconstructed from its multiscale edges. The algorithm using only multiscale edges reduced the computation complexity in the fusion process. The fusion rule enabled the fusion image to hold as much as edge in formation of original images. Furthermore, the algorithm could compress data even without using any compression coding method, which reduced memory cost and decre ased data transfer bandwidth. The experimental results demonstrate that fusion i mages generated by this algorithm contain the information of original images eff ectively.