Comparative Analysis of Medical Images Denoising using Wavelet, Curvelet and Shearlet Transforms

摘要

The major issue that arises in medical imaging is the difficulty of transmitting large volume of medical data with low bandwidth. But these Medical images during acquisition acquire different types of noise and so it is necessary to denoise them. Wavelets and its variants such as Curvelets, Shearlets can effectively denoise medical images. Any denoising system performs three operations: transformation, thresholding and inverse transformation. Transformation packs as much information as possible into smaller no. of transform coefficients. Thresholding means neglecting certain coefficients. For doing this we have to decide the value of threshold. A comparative analysis is drawn between MRI denoising of different parts of the body with added Gaussian noise using Wavelets, Curvelets and Shearlets. Thresholding is the most critical step in denoising. In this paper we have modified the existing thresholding techniques to denoise the images effectively. Various quantitative measures such as Peak Signal to Noise ratio, Mean Square error, Maximum Difference, Normalized Cross Correlation etc. are used to measure the effectiveness of the proposed algorithm. Shearlet transform is the most effective in removing Gaussian noise followed by Curvelet transform and Wavelet transform.