A Comparative study of a Face Components based Model of Ethnic Classification using Gabor Filters

The determination of ethnicity of an individual can be very useful in face recognition and person identification. In this paper, we propose a model of ethnic classification of a person from face images. The proposed model detects facial landmarks such as eyes, nose and mouth in a face image and then applies Gabor filters to each component to extract key facial features. K-Means, Naive Bayesian(NB), Multilayer Perceptron(MLP), SVM(Support Vector Machines) are then used to classify the human face image into different ethnic groups. Classification is performed in 2-class ( Asian and Non- Asian), 3-class (Asian,White and Black) or 4-class (Asian, Indian, White and Black). The results show that the mouth and the nose outperform the eyes in the characterization of ethnicity, and the classification is improved when all these components are used. It is also important to note that although few face components are used, the proposed model is comparable or even outperforms some existing models