Impulse noise filtering based on noise-free pixels using genetic programming

Generally, the impulse noise filtering schemes use all pixels within a neighborhood and increase the size of neighborhood with the increase in noise density. However, the estimate from all pixels within neighborhood may not be accurate. Moreover, the larger window may remove edges and fine details as well. In contrast, we propose a novel impulse noise removal scheme that emphasizes on few noise-free pixels and small neighborhood. The proposed scheme searches noise-free pixels within a small neighborhood. If at least three pixels are not found, then the noisy pixel is left unchanged in current iteration. This iterative process continues until all noisy pixels are replaced with estimated values. In order to estimate the optimal value of the noisy pixel, genetic programming-based estimator is developed. The estimator (function) is composed of useful pixel information and arithmetic functions. Experimental results show that the proposed scheme is capable of removing impulse noise effectively while preserving the fine image details. Especially, our approach has shown effectiveness against high impulse noise density.