This paper presents an improved algorithm for suppressing the synthetic aperture radar image speckle noise

in which the double density dual tree complex wavelet transform (DD-DT CWT) was combined with the bivariate shrinkage function(BSF) with local variance estimation.The SAR image was firstly decomposed by the DD-DT CWT

the bivariate probability density function was used as statistic correlation model for wavelet coefficients and their parent

and the corresponding bivariate shrinkage function was obtained by Bayes' estimation theory.Then the wavelet coefficients were shrunk by the BSF

in which different orientations were nonlinear processed adaptively

finally the denoised image was reconstructed by all the update coefficients.The algorithm presented was tested by simulated and actual SAR image

and compared with other methods.Results are presented to verify that

the new algorithm significantly outperforms the traditional wavelet-transform-based denoising algorithm

and effectively denoises the noisy image as well as preserves the particulars.The performance index of the denoised image were simultaneously improved.