Mongoloid and Non-Mongoloid Race Classification from Face Image Using Local Binary Pattern Feature Extractions

Abstract

One of the areas on the human body that has the most dominant racial trait is the face. This research build the classification system for Mongoloid and non-Mongoloid race based on the area in the periorbital area of facial image. We use Local Binary Pattern to extract texture features on periorbital facial area. To classify the LBP features, we use Support Vector Machine classifier. The accuracy obtained from the system is 99.38%.