A hybrid genetic algorithm for color image denoising

In this paper, we address the problem of impulsive noise reduction in color images through an evolutionary approach. We designed a new hybrid genetic algorithm, called GARIN, which takes as input a noisy image and generates as output a reduced noise version of the same image. As part of its evolutionary process, GARIN integrates the execution of robust and adaptive filters from literature with the aim to combine and improve their results. Unlike most genetic algorithms designed for color image denoising, our algorithm evolves a set of images instead of the set of parameters of the filters. Experimental results show that, compared with other filters of literature, the algorithm GARIN efficiently removes impulsive noise in color images while, at the same time, it preserves their main features.