Robust trimmed mean filter for color impulse noise enhancement

Abstract

There was different type analysis work already proposed in the past for the enhancement of gray scale images these are corrupted by salt and pepper noise, however in case of color image denoising for color impulse noise need more effort during this field. This research object to epitomize a powerful approach for the removal as well as enhancement of color and gray level images those are extremely corrupted color impulse noise also on gray scale impulse noise. In color impulse noise images are corrupted by the pixel value black and white are making certain that red, green, blue, black and white dot within the image. In proposing algorithm take a color noise image separated into 3 subparts, i.e. red, green and blue color pixel matrices, then these 3 matrices are checked for noisy or noisy free pixels. Then choose 3×3 size window of every frame with targeted pixel value check if its value is zeros or two fifty five then the image is corrupted otherwise noise free image. If the image is noisy and target pixels neighboring pixel value is between 0's and 255's currently we interchange values of the corrupted pixel by unsymmetrical trimmed median and if target pixels neighboring pixel value is 0's or 255's now we interchange the value of pixels by the unsymmetrical trimmed average. The proposed filter algorithm shows higher simulation results as compared the existing algorithms. The simulation result shows efficient performance in expressions of peak signal to noise ratio (PSNR) and mean square error (MSE) and complexity additionally shows higher end up in terms of human and computer vision.