Application of genetic algorithm in median filtering

Abstract

Images are often corrupted by impulse noise due to errors generated in noisy sensors or communication channels. Two types of impulse noise can be defined: 1) fixed-valued and 2) random-valued. In many applications it is very important to remove noise in the images before some subsequent processing such as edge detection, object recognition and image segmentation. In this paper, an adaptive filtering using genetic algorithm is proposed. In the simulations over various images, the proposed partition based median (PBM) filter using genetic algorithm in training has demonstrated better results in noise suppressing than competitive filters based on median filtering in terms of SNR (dB) as well as the perceived image quality. The proposed filter outperforms other median based filters in removing different types of noise: impulse noise (fixed-valued and random-valued), Gaussian noise and mixed Gaussian and impulse noise.