Image denoising using a multivariate shrinkage function in the curvelet domain

A new method based on the curvelet transform is proposed for image denoising. This method exploits a multivariate generalized spherically contoured exponential (GSCE) probability density function to model neighboring curvelet coefficients. Based on the multivariate probability model, which takes account of the dependency between the estimated curvelet coefficients and their neighbors, a multivariate shrinkage function for image denoising is derived by maximum a posteriori ( MAP) estimator. Experimental results show that the proposed method obtains better performance than the existing curvelet-based image denoising method.