ICTACT Journal on Image and Video Processing

K. Vasanth ¹, S. Karthik ²

Affiliations
1 Sathyabama University, Tamil Nadu, IN
2 Cognizant Technology solutions, Tamil Nadu, IN

Abstract

The proposed work aims in the restoration of images corrupted by Gaussian noise, impulse noise. The new algorithm significantly removes different noises and produce better image quality than standard median filter (SMF), Centre weighted median filter (CWF) and threshold decomposition filter (TDF).The proposed algorithm (PA) is tested on different images corrupted by all two noises and is found to produce better results in terms of the qualitative and quantitative measures of the image for noise densities up to 30% noise level for impulse noise, mean zero and 0.9% variance of Gaussian noise. The filter works well for speckle noise up to 0.8% variance.

Keywords

Impulse Noise, Median Filter, Threshold Decomposition, Non-Linear Filter.

Full Text

     

K. Vasanth ¹, S. Karthik ²

Affiliations
1 Sathyabama University, Chennai, Tamilnadu, IN
2 Cognizant Technology solutions, Chennai, Tamilnadu, IN

Abstract

A novel algorithm is proposed for the restoration of images corrupted by impulse noise; this work aims at a novel filter whose window size is fixed (3X3) for all noise densities. The new algorithm significantly produces better image quality than standard median filter (SMF), adaptive median filters (AMF), and center weighted median filter (CWMF) and threshold decomposition filter (TDF). Unlike the other filters, the proposed algorithm computes median if and only if there is a corrupted pixel and replaces it by the median value, if the median is corrupted then average of uncorrupted pixels in the current processing window is replaced else preprocessed pixel value is replaced. The proposed method removes the noise effectively even at noise level as high as 85% and preserves the edges without any loss up to 80% of noise level. The proposed algorithm (PA) is tested on different images and is found to produce better results in terms of the qualitative and quantitative measures.

Keywords

Impulse Noise, Median Filter, Adaptive Median Filter, Decision Based Filter, Weighted Filter.

Full Text