A new speckle reduction algorithm based on subband decomposition is presented. The subband decomposition is performed using a DFT/RDFT-based symmetric, zero-phase, nonoverlapping digital filter bank. The SAR image is decomposed using the proposed filter bank structure and a threshold value is estimated according to noise variance in each subband, and used for soft thresholding. The despeckled image is obtained from the thresholded subband coefficients by a new perfect reconstruction algorithm. Experimental results show that the proposed method has competitive noise removal and detail preservation properties as compared to conventional approaches such as median filtering and wavelet transform based methods.