A new image transform,namely,the contourlet transform is introduced,which can capture the intrinsic geometrical structure of image. Furthermore,a novel image denoising scheme based on contourlet is presented. Via calculating variance homogeneous measurement (VHM),the locally adaptive window is determined to estimate the shrinkage factor optimally,then the contourlet coefficient is shrunk using the shrinkage factor. The scheme utilizing the correlation of contourlet coefficients in the same subband along the edge or contour of the image,which can get the tradeoff between “noises removing”and “details preserving”.In numerical comparisons with various methods,the presented scheme outperforms the traditional contourlet denoising method based on hard-thresholding and Wiener filter in terms of PSNR. Experiments also show that this scheme could not only remove the noises effectively,but also suit for the neutron radiography system.