Study on Human Recognition by Ear Image Based on Genetic Local Search

Abstract

An individual recognition system using the genetic local search with respect to ear images has been examined. We generated a set of individuals which are ramdomly selected from the ear images consisted of 50 registrants, and carried out genetic operations to the set of the individuals in order to lead to new generation. The local search is applied to some top elites with the best fitnesses in the new generation. The best fitness is searched between some positions selected near the elite position. Individual recognition experiment using the ear image with the uncertain data was carried out in respect of the method. As the result, it is noted that the registrant recognition rate was 100 percent, and it became 100 percent, at the un-registrant denial rate.