A new speckle reduction algorithm based on subband decomposition and multiscale edge detection is presented. The subband decomposition is performed using a DFT/RDFT-based symmetric, zero-phase, nonoverlapping digital filter bank. The noisy images are decomposed into sub-images using RDFT transform based subband decomposition, an edge image is obtained for each level using these sub-images, then a threshold value is estimated according to noise variance in each subband and soft-thresholding is applied on the non-edge regions of the subband images. The despeckled image is obtained from the thresholded subband images using the inverse RDFT transform. The proposed speckle reduction method is applied to several test images corrupted by speckle noise as well as experimental RADARSAT/SAR images. Despeckled images show that this method has better performance than the case where no edge information is used.