Abstract: A new wavelet transform-based edge detection algorithm for images is presented in the paper. The processes of the algorithm are as follows: the multiscale dyadic wavelet transforms are first performed on the image along horizontal, vertical and diagonal directions to obtain wavelet coefficients; the method of adjacent scales wavelet coefficients multiplication is then applied to remove noise and extract the maximum points of the wavelet coefficients product; the maximum points are finaly merged to conform the image edges. The result of emulation experiment indicates that the algorithm has the ability of controlling noise and detecting edges effectively. The detection results are superior to the traditional detection results.