An image denoising method based on fast discrete curvelet transform and Total Variation
Hongzhi Wang, Liying Qian, Jingtao Zhao · 2010
In this paper, A new hybrid image denoising method is proposed based on curvelet transform and Total Variation(TV) algorithm for removal of addictive white Gaussian noise. Firstly, perform fast discrete curvelet transform using USFFT to the noisy image, then perform hard threshold to the curvelet coefficients of every sub-band and reconstruct the modified coefficients to obtain primary denoised image. In order to remove the surround effect brought by curvelet transform, we conduct further filtering by TV method with about 10 iterations only. The experiment results show that the hybrid algorithm suppress surround effect without appearing staircase effect of TV method effectively. We obtain better visual quality and PSNR comparing to the curvelet transform based method. At the same time, the proposed method takes less computational time than TV filter and achieves better synthesized performance.