A Spatially Adaptive Denoising Algorithm Based on Curvelet Transform
A new approach for image denoising based on the Curvelet transform is presented in this paper.The existing theory for Curvelet and Ridgelet suggests that these new approaches can outperform wavelet method in certain image processing including image denoising,edge detection and image enhancement.However in original digital Curvelet transform it uses a simple hard-thresholding rule for filtering the noisy Curvelet coefficient which of course causes some problems such as killing too many signal Curvelet coefficients that might contain useful image information.Here we (luce) BayesShrink denoising scheme into Curvelet domain (is) an adaptive,data-driven thresholding approach for image denoising,namely CurShrink.The threshold is derived in a Bayesian framework,and the prior used on the Curvelet coefficients is the generalized Gaussian distribution (GGD)widely used in image processing applications.The approach is valid because Curvelet transform produce correlated Curvelet coefficients,along the edge or curve of the image; the large Curvelet coefficients maybe have large Curvelet coefficients as it neighbors.Experimental results show that the proposed method is better than hard-thresholding denoising scheme in wavelet and curvelet domain.
Curvelet Transform image denoising Multi-scale Geometric Analysis(MGA) CurShrink
Peng Feng Feng Yang Biao Wei Deling Mi
Key Laboratory of Opto-electronics Technology & System Ministry of Education Chongqing University Chongqing, China, 400044
国际会议
沈阳
英文
685-689
2012-07-27（万方平台首次上网日期，不代表论文的发表时间）