Reducing Image Noises Using Genetic Algorithm's Uniform Crossover

Abstract

Images could display visual information more than those of text data. However, when transmitted and acquired through communication channels, those images are always spoiled with noises that will reduce the quality of the image. Noisy image could not provide good quality image for further image processing due to poor quality. In image processing, standard genetic algorithm steps could be used to enhance image quality. The purpose of this research is to deploy uniform crossover of genetic algorithm to reduce noise in order to produce better offsprings. In every noise type, the obtained value of Mean Square Error (MSE) and Peak Signal-to-Noise Ratio (PSNR) resulted in image noise reduction were calculated and analyzed to see how both values of MSE and PSNR in average will change. For this purpose, we conducted tests with Pc values of 0.2, 0.4, 0.6, and 0.8, each with 100, 200, 300, 400, 500, and 1000 maximum number of generations, respectively. Result shows that uniform crossover obtained the best performance in reducing erlang noise and the worst performance in reducing localvar noise on three categories of images.