Open AccessAn Effective Noise Adaptive Median Filter for Removing High Density Impulse Noises in Color Images
Author(s) -
Salman Saleem,
T. Abdul Razak
Publication year - 2016
Publication title -
international journal of electrical and computer engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.277
H-Index - 22
ISSN - 2088-8708
DOI - 10.11591/ijece.v6i2.pp611-620
Subject(s) - impulse noise , computer science , median filter , artificial intelligence , computer vision , image quality , peak signal to noise ratio , adaptive filter , impulse (physics) , filter (signal processing) , pattern recognition (psychology) , image processing , algorithm , image (mathematics) , pixel , physics , quantum mechanics
Images are normally degraded by some form of impulse noises during the acquisition, transmission and storage in the physical media. Most of the real time applications usually require bright and clear images, hence distorted or degraded images need to be processed to enhance easy identification of image details and further works on the image. In this paper we have analyzed and tested the number of existing median filtering algorithms and their limitations. As a result we have proposed a new effective noise adaptive median filtering algorithm, which removes the impulse noises in the color images while preserving the image details and enhancing the image quality. The proposed method is a spatial domain approach and uses the 3×3 overlapping window to filter the signal based on the correct selection of neighborhood values to obtain the effective median per window. The performance of the proposed effective median filter has been evaluated using MATLAB, simulations on a both gray scale and color images that have been subjected to high density of corruption up to 90% with impulse noises. The results expose the effectiveness of our proposed algorithm when compared with the quantitative image metrics such as PSNR, MSE, RMSE, IEF, Time and SSIM of existing standard and adaptive median filtering algorithms.