Normal Inverse Gaussian Model-Based Image Denoising in the NSCT Domain

Jian Jia,Zhihua Zhao,Yongxin Zhang,Li Chen

doi:10.1155/2015/851313

Abstract

The objective of image denoising is to retain useful details while removing as much noise as possible to recover an original image from its noisy version. This paper proposes a novel normal inverse Gaussian (NIG) model-based method that uses a Bayesian estimator to carry out image denoising in the nonsubsampled contourlet transform (NSCT) domain. In the proposed method, the NIG model is first used to describe the distributions of the image transform coefficients of each subband in the NSCT domain. Then, the corresponding threshold function is derived from the model using Bayesian maximuma posterioriprobability estimation theory. Finally, optimal linear interpolation thresholding algorithm (OLI-Shrink) is employed to guarantee a gentler thresholding effect. The results of comparative experiments conducted indicate that the denoising performance of our proposed method in terms of peak signal-to-noise ratio is superior to that of several state-of-the-art methods, including BLS-GSM, K-SVD, BivShrink, and BM3D. Further, the proposed method achieves structural similarity (SSIM) index values that are comparable to those of the block-matching 3D transformation (BM3D) method.

Highlights

The objective of image denoising, a classical but still very active area in image processing, is to retain useful details while removing as much noise as possible to recover the original image from a noisy image [1]
This paper proposes a novel normal inverse Gaussian (NIG) model-based method that uses a Bayesian estimator to carry out image denoising in the nonsubsampled contourlet transform (NSCT) domain
The images were obtained from http://www.cs.tut.fi/ ∼foi/GCF-block-matching 3D transformation (BM3D), and we verified that they were the same as the images used by Dabov et al [15]

Summary

Introduction

The objective of image denoising, a classical but still very active area in image processing, is to retain useful details while removing as much noise as possible to recover the original image from a noisy image [1]. This paper combines current multiscale, multiresolution analysis to propose a novel nonsubsampled contourlet transform (NSCT) denoising scheme based on the normal inverse Gaussian (NIG) probability density function (PDF). Many researchers have begun to focus on methods of expressing image geometry structures more effectively in order to overcome the inability of wavelet transformation to adequately represent the geometric structure information of images These activities have led to the introduction of ridgelet transform [18], curvelet transform [19], and contourlet transform [20] for singular analysis of two-dimensional or higher images, techniques which can achieve good sparsity for spatially localised details, such as edges and singularities. A relatively large threshold value may destroy the high-frequency information of the image and produce a false Gibbs phenomenon in the denoised image

Marginal Statistical Modelling on the NSCT Subband Coefficients

Monte-Carlo Estimates of NSCT Domain Noise Coefficient

Experimental Results

Method

Conclusion