Generating art tile patterns using genetic algorithm

Generating art, using artificial intelligence techniques, is a new and fantastic research area. Artificial intelligence represents many techniques which can be used to achieve this goal. In particular, evolutionary computing plays an important role to generating visual art and music. Genetic algorithm is specially a very useful means which can be used to create art products. However, it is very difficult to determine several proper criteria to evaluate fitness of art products. Furthermore, the definition of quantitative metrics based on qualitative criteria, is still a very challenging research area. In this paper, experimentally, a novel application is implemented and introduced, that produces art tile patterns using genetic algorithm. The horizontally and vertically consecution of gray levels of pixels, plus having an equalized histogram, have been considered as basic criteria to construct the fitness function. Also the symmetric nature of tile patterns, have been used perfectly and only a quarter of tile pattern have been considered as a chromosome. The proposed approach starts with one population, uses reproduction and mutation, and selects the best offspring in iterations. Finally according to the visual results showed in this paper, we can conclude the efficiency of genetic algorithm to generate art tile patterns.