New image denoising algorithm based on the second generation curvelet transform

Compared to the wavelet transform the curvelet transform is more superior in expressing curve singular function anisotropy and image edge.In view of the insufficiency of wavelet image noise reduction,the characteristics of the fast discrete curvelet transform based on the wrapping algorithm is analyzed,and a new denoising method combining with cycle shift,Ell iterative and Monte Carlo methods' threshold rule is proposed.This algorithm fully uses the correlation of curvelet coefficient to eliminate the scratch distortion and the ringing effect caused by lack of image translation invariability of the Curvelet transform.Experimental results show that the algorithm yields denoised images with higher PSNR and better visual effects in comparison with traditional wavelet denoising,second-generation wavelet denoising,wavelet packet denoising and hard threshold denosing based on curvelet transform.