Adaptive Threshold and Directional Weighted Median Filter-Based Impulse Noise Removal Method for Images

Abstract

Elimination of impulse noise in image snap shots with side renovation is one of the complex duties in digital image processing. In this paper, the removal of random impulse noise is done in two important levels. In first level, the detection of the impulse noise is done on the premise of a double threshold selecting strategy after which in the another level, elimination of impulse noise is done by the usage of median filter and directional weighted median filter relying upon the noise map (Nmap) construction of corrupted pixels detected within the first level. The proposed method makes use of the statistical characteristics of noisy image graphs and the brink obtained is adaptable to one of a kind of snap shots and noise conditions. Comparative evaluation with different widespread de-noising techniques shows that the proposed method outperforms in terms of PSNR, SSIM, NMSE and Computation Time (CT) of the distinct trying out test images, with exclusive noise levels.