Adaptive Thresholds Algorithm of Image Denoising Based on Nonsubsampled Contourlet Transform

In this paper, we propose an adaptive threshold estimation method for image denoising based on nonsubsampled contourlet transform (NSCT). The NSCT is developed because of the lack of translation invariance of the contourlet transform. It is good at isolating the smoothness along the contours and can overcome the Gibbs-like phenomena around singularities. The proposed algorithm can adapt different thresholds on different scales and different directions depend upon the subband size and number of decompositions. After determining the thresholds, we adopt a new thresholding function which is continuous around the threshold and reduces the deviation between the coefficients and the original ones. The experiments show that the proposed approach can improve the SNR values, especially for the images that include mostly contours.