ISMSIT 2017, Tokat, Turkey, 2 - 04 November 2017, vol.1, no.1, pp.9-14
In this study, Optimization of myriad filters by genetic algorithm from evolutionary algorithms are studied and this performance is tested for α-stable noise situations in different characteristics. Also performance of Genetic Algorithm is compared with a derivative based classical algorithm. Myriad filter structure has been used in recent years as a powerful nonlinear filter structure for impulsive noisy environments (especially α-stable noise). This filter structure has been successfully applied to the fields of communication, signal and image processing. The α-stable noise depends on the change of α in the form 0 <α ≤ 2. In this case, the noise that occurs in case of α = 1 is Cauchy and the noise that occurs in case of α = 2 is Gaussian distributed. Evolutionary algorithms are frequently used to solve different problems due to learning, generalization, application of different problems easily and noise tolerance. As a result of the studies made, for the α-stable noise cases, it has been observed that Myriad filter weights can be detected with low optimization error by using Genetic algorithms. When evaluated generally, Genetic algorithm has been found to be successful in optimizing myriad filters.