To diagnose the brain cancer, Magnetic Resonance Image is considered as one of the efficient technologies. For disease diagnosing, the information received from MRI images is used by the radiographer to provide the concern treatment for the patient further. However, noise is always contained in MRI images. From images, elimination of noise is very critical and important task. In prior to the proceeding to next task, it’s required to be implement the preprocessing steps if in case the images are accumulated with the noise. For removing of noise, filtering algorithms are used. A major drawback in the existing denoising algorithms is that the resultant image suffers from blurring as the edges are affected in the process of filtering. This paper proposes an image denoising technique with the spectral total variation which reduces the noise in the image without blurring the edges. The existing algorithms have been implemented to compare the performance of the proposed system.