Gaussian Denoising By Time Domain and Frequency Domain Filters for MRI Brain Images

Abstract

Image restoration is the study of recovering the undistorted image. In this study, Time Domain (TD) filters and Frequency Domain (FD) filters are designed to discover how good a restoration of Magnetic Resonance Image (MRI) of brain is possible. The main property of image restoration by filters is that the edges in the image should be preserved while denoising the image. At first, the Gaussian noise is generated with zero mean and different variances. Then, the system generated Gaussian noise is added to the 256 by 256 pixel MRI brain images in the REpository of Molecular BRAin Neoplasia DaTa (REMBRANDT) database images. Then the noisy images are denoised by the TD and FD filters. The results of processing images using FD filter show an improvement over the TD filters in terms of Structural Similarity Index Measure (SSIM) and Peak Signal to Noise Ratio (PSNR). This is because the FD filter provides the details of high frequency low frequency components effectively than TD filters.