A Noise Fading Technique for Images Highly Corrupted with Impulse Noise

Abstract

A novel noise fading technique based on noise detection and median filtering is proposed in this paper. This technique can be used for denoising the images extremely corrupted with impulse noise. This paper introduces a spike detection technique (SDT) and pixel restoring median filter (PRMF) for denoising the corrupted images. The SDT is used for discriminating between corrupted and uncorrupted image pixels. The corrupted pixels are restored using PRMF technique. Our iterative denoising technique is repeated until the corrupted pixels in the recovered image reduce to zero. The performance of our denoising scheme is evaluated with salt and pepper noise and also with random impulse noise for different standard images. It is observed that the proposed denoising scheme outperforms all existing impulse-denoising schemes. This technique can also be used for color image impulse noise removal. This technique can remove very high noise up to 98% and the images denoised with our method shows improvement in terms of visual quality, PSNR value and mutual information. This scheme prevents image blurring and is computationally simple. Hence it is suitable for real-time applications