Feature Selection for Character Recognition Using Genetic Algorithm

We propose a novel method of feature selection for character recognition using genetic algorithms (GA). The feature is assigned to the chromosome, and values of "1" and "0" are given to the chromosome; corresponding to features that are respectively used and unused for recognition. GA decreases the number of chromosomes which take the value of "1" while changing generations. The proposed method selects only genes for which the recognition rate of training samples exceeds the predetermined threshold as a candidate of the parent gene and adopts a reduction ratio in the number of features used for recognition as the fitness value. Consequently, it becomes possible to reduce the number of features while maintaining the recognition rate. On the experiment for similar-shaped character recognition, the proposed method achieved a higher recognition rate and larger decrease of the number of features compared with Fisher\´s criterion.