Anisotropic Diffusion Based Unsharp Masking and Crispening for Denoising and Enhancement of MRI Images

1. Abstract-Unsharp masking and crispening of Magnetic Resonance Images (MRIs) is a vital factor for good visual inspection, accurate parameter estimation, and for further image processes like feature extraction and classification. To accomplish this objective, a method is proposed that is adapted from anisotropic diffusion filter(ADF). It utilizes ADF regularization as an intermediate step for smoothening of MRI image. It is an expanded and improved version of “Perona and Malik” Diffusion which gives a better image quality assessment when compares to other methods. The image quality assessment criterions like Peak signal-to-noise ratio (PSNR), mean square error (MSE) and structural similarity index (SSIM) are determined with respect to full reference image. Some no reference metric like Perception based Image Quality Evaluator (PIQE) and Blind/Reference less Image Spatial Quality Evaluator (BRISQUE) are also determined. The corresponding results of full reference image for kaggle data-set. are estimated as 39.13dB, 7.21 and 0.99 respectively. The average result regarding for no reference metric are given as 48.84 and 37.13 respectively. The following outline has been executed in MATLAB R2015b which is used for MRI images of various resolutions, shape and size. The achieved outcomes expressed that overall image is deniosed & enhanced by the proposed method.