Artificial Neural Network for Prediction of Ethnicity Based on Iris Texture
Anahita Zarei, Duxing Mou · 2012
Iris is a reliable biometric that is unique, remarkably stable through the life of an individual, and easy to capture. Many applications include verifying the identity of a subject or identifying an unknown individual from a list of possibilities that involve searching a large database. Prediction of ethnicity separates the data into subcategories that will make this search much faster. We have proposed a new approach, namely artificial neural network to predict the ethnicity based on iris texture features. We have achieved the correct classification rate of 93.3% Caucasian/Asian on a person-disjoint set. This result outperforms the previous works using statistical methods on the same database.