A Continuous Decision Based Multi Kernel Median Filter For Noise Removal on Brain MRI Images

Abstract

Impulse noises in images are caused by bit errors in transmission and signal acquisition.  Salt and pepper noise and random noise are also known as Impulse Noise.    As per the statistical analysis of noise in Brain MRI image shows salt and pepper noise is one of the most common which affect the accuracy of the tumor detection.  Many nonlinear algorithms have been proposed to remove salt and pepper noise. But without damaging the edges is the difficult Task. Noise removal without damaging the edges is proposed in this paper.   If the noise density increases, the effectiveness of the filter will be decreased. This is the major drawback of the existing algorithms.  This paper discusses many noise removal techniques and proposes a novel noise removal technique using Continuous Decision Based Multi Kernel Median Filter (CDBMKMF). The proposed CDBMKMF algorithm attempts to eliminate noise in high noise density images with better  PSNR values.   Image pixels are checked for the occurrence of salt and pepper noise and removed effectively.  Using Mean Square Error (MSE) and Peak Signal to Noise Ratio (PSNR), the proposed methods are validated and compared with existing algorithms.  This paper also evaluates the proposed algorithm with standard and unsymmetrical median filters.