Diversity in Crowding Selection to Improve Fractal Coding Technique Performance

Abstract

Since Barnsley conception of fractal coding, this research topic was rapidly grown and considered as the most promising technique. The high encoding time prevents the popularity of this technique. Crowding as an improved Genetic algorithm (GA) is used for fractal coding method to improve the searching computational time between range- domain blocks. Although, crowding method is one of the approaches that maintain population diversity, but still the offspring replaces the most similar individual. In this method, each offspring is completed with one of its parents and replaced it. That means, crowding method concentrated on best element’s population search space, which makes the new generation influenced by the properties of the previous generation. To support global exploration and prevent trapping in local optima, a new probability is added, which helps to satisfy diversity in the population selection from both the best and worst individuals. This prevents the algorithm from being trapped in local optima. By this method, the time complexity of the algorithm is improved because it makes a balance between reaching the optimal solution and population diversity.