Abstract— The morphological Mean Filter method is used to restore image corrupted by high-density impulse noise. A new method for de-noising is developed for MR (Magnetic Resonance) images which have Rician noise. This method is an enhancement of the Morphological Mean Filter (MMF). In this method, first Grey level Co-occurrence Matrix (GLCM) method is applied to noisy images and then the MMF method is applied to remove noise from the image. Experimental results show that the Proposed method effectively removes the noise as compared to the MMF method. Index Terms—GLCM, impulse noise, morphological mean filter, noise removal, Rician noise. I. INTRODUCTION Image denoising or Noise removal can be described as a procedure which is used for the removal of noise from a picture. The noise corrupts all the features of a picture during the acquisition procedure or transmission. The image denoising process maintains the quality of a picture. In the medical field, timely recognition of disease is necessary. The random noise affects the quality of images during image acquisition step. This results in unwanted outcomes and bad optical quality of a picture which lowers the visibility of low contrast objects. The extraction of concealed details, image data and recovery of fine details is essential for the removal of noise in the applications of medical imaging. These noise corrupted MR (Magnetic Resonance) images influence the medical diagnosis procedure. In general, various techniques are used for eliminating noise from an image. For denoising MR images, various algorithms are proposed earlier. In a good quality denoising tool, the noise suppression process should not affect the quality of an image and this process should not deteriorate the useful features of an image as well [1]. In the MR images, the boundaries are very important; therefore, the preservation of boundaries is very necessary for the denoising process. These days different scanning techniques are used in numerous applications. These techniques improve spatial resolution, SNR, and acquisition speed. However, at the time of analysis, the noise present impacts the analytic and visual quality of MRI. There are several kinds of noises included in images. Among these, few commonly known noises are explained below: Gaussian Noise: The image includes this noise at the time of image acquisition. For instance, the transmission noise, Manuscript received January 9, 2021; revised April 19, 2021. Jyoti Choudhary and Alka Choudhary are with J. C. Bose University of Science and Technology, YMCA, Faridabad, Haryana, India (e-mail: jyoti.attri11@gmail.com, alka.attri@gmail.com). circuit noise, sensor noise, etc. caused due to low light. The spatial filtering is applied such that the noise present in the color image can be removed [2]. A noise has Probability density function [PDF] of normal distribution. This function is called Gaussian distribution as well. Impulse Noise: Salt & Pepper noise or Spike noise are some other names given to the impulse noise. The malfunctioning of pixels within the camera sensors, usage of the noisy channel for communication, or the existence of faulty locations within the memory makes the image noisy. The image pixels do not connect to this type of noise. Rician Noise: Due to the presence of this noise, a bias is included in the color measurements. Thus, there is an extensive alteration of the shapes and orientation within the diffusion tensor magnetic resonance images. Thus, due to the presence of Rician noise in images, huge impacts are caused by the attributes of images. II. RELATED WORK In [3], the author compared outcomes of several earlier proposed mean & median filtration techniques and proposed a new method in conjunction with spatial adaptive masking filtration to remove impulse noise. In [4], depending upon the morphological and fractal techniques, 3D MR image was designed using segmentation and denoising approaches. The paper [5] proposed MCDnCNN model which removed the Rician noise from 3D MR images. The authors [6] developed edge-preserving denoising technique in the field of medical image processing. The various denoising techniques using Fusion of local and non-local filter [7], using Contourlet Transform and Threshold Shrinkages Techniques [8], using wavelets [9]-[11], non-local mean filter [12], and other techniques for denoising [13], [14] and enhancing the image [15]-[17] were proposed. III. PROPOSED ALGORITHM The proposed algorithm is used to remove Rician Noise from Magnetic Resonance Images (MRI). The proposed algorithm is an enhancement of Morphological Mean Filter [18] which is used to remove impulse noise from images. The proposed filter consists of two modules shown in Fig. 1: Feature extraction module and Noise filtering module. In the first module, Feature extraction module uses GLCM (Grey level Co-occurrence Matrix) [19] algorithm to calculate some features like similarity features, contrast factors, etc. These features are given as input to the second module, Noise filtering Module. The Noise filtering module uses the MMF (Morphological Mean Filter) [18] to de-noise the image. Enhancement in Morphological Mean Filter for Image Denoising Using GLCM Algorithm Jyoti Choudhary and Alka Choudhary International Journal of Computer Theory and Engineering, Vol. 13, No. 4, November 2021 134 DOI: 10.7763/IJCTE.2021.V13.1302