Image denoising in contourlet domain based on a normal inverse Gaussian prior

This paper presents a new image denoising algorithm based on the modeling of contourlet coefficients in each subband with a normal inverse Gaussian (NIG) probability density function (PDF). This PDF is able to model the heavy-tailed nature of contourlet coefficients and the local parameters model the intrascale dependency between the coefficients. Within this framework, we describe a novel method for image denoising based on designing maximum a posteriori (MAP). Furthermore, the cycle spinning algorithm is employed to modify the Gibbs phenomenon around edges caused by the lack of translation invariance of the contourlet transform. Experimental results prove that the new method can remove Gaussian white noise effectively, reserve image edges better and enhance the peak signal-to-noise ratio.