Interactive genetic algorithm-aided generation of carpet pattern

摘要

Pattern generation for carpets is an expensive and time-consuming task that requires skills, imagination, and creativity. In recent years, interactive genetic algorithm (IGA) has been successfully utilized in generating patterns in different fields such as architecture, graphic art, and music among many others. In these applications, fitness functions cannot mathematically be defined, and the fitness value of each pattern combination is required to be evaluated by designer. Therefore, IGA can be utilized to serve as an effective tool to accelerate the design process. In the present work, a modified IGA is developed to generate patterns for machine-made carpets. Each pattern consists of a set of basic elements. Some basic pattern elements are collected, categorized, and coded as gene values. In each generation, new population is formed on the basis of observer's evaluation. All generated patterns are simultaneously shown for observer to assign a fitness value to each one. Hence, a new population can be achieved using the genetic algorithm. Our results show that acceptable patterns can be obtained after a limited number of generations. The system is designed in such a way that it can be used by unprofessional users.