Iris Texture Analysis for Ethnicity Classification Using Self-Organizing Feature Maps

Ethnicity Classification from iris texture is a notable research in the field of pattern recognition that differentiates groups of people as distinct community by certain characteristics and attributes. Several ethnicity classification systems have been developed using Supervised Artificial Neural Network and Machine Learning algorithms. However, these systems are limited in their clustering ability and require prior definition of image classes which lowers its classification rate. Therefore, this work classified iris images from Nigeria, China and Hong Kong origin using Self-Organizing Feature Maps (SOFM) blended with Principal Component Analysis (PCA) based Feature extraction. Left and right irises of 240 subjects constituting 480 images were acquired online from CUIRIS (Nigeria), CASIA (China) and CUHK (Hong Kong) datasets, and normalized to a uniform size of 250 by 250 pixels. Three hundred and thirty six (336) images were used for training while the remaining 144 were used for testing. The system was implemented in Matrix Laboratory 8.1 (R2013a). The performance of the classification system was evaluated at varying thresholds (0.2, 0.4, 0.6 and 0.8) and 93.75% Correct Classification Rate (CCR) was obtained.