Algoritme Genetika untuk Optimasi Pembobotan Fitur pada Temu Kembali Citra

Abstract

Content based image retrieval is developed to retrieve image based on color, shape, and texture feature. One of the problems to be solved in content based image retrieval is how to combine these features to get the optimum retrieval result. Combining multiple features in image retrieval process can be implemented using feature weight assignment. One of feature weight assignment methods is each feature weight is assigned by hand. However this approach is subjective and lack of theory foundation. In this research an automatic feature weight assignment approach based on genetic algorithm is proposed. Genetic algorithm is used for optimizing feature weight in order to get the optimum retrieval result. Image features is used in this research are color, shape, and texture feature. In genetic algorithm, a chromosom represented by features weight. These chromosomes get into genetic operator process such as selection, crossover, and mutation until maximum generation is reached. Result of genetic algorithm process get an optimum feature weight. It used for image retrieval process. The retrieval evaluation used recall precision graphic. The experiment showed feature weight assignment based on genetic algorithm can improve average precision in image retrieval.