Iris recognition from distant images based on multiple feature descriptors and classifiers

This paper proposes a new approach to recognize iris from distantly acquired facial images by utilizing multiple feature descriptors and classifiers. Firstly, Log-Gabor (LG), Contourlet Transform (CT), Gradient Local Auto-Correlation (GLAC) and Convolutional Neural Network (CNN) descriptors are employed on segmented normalized iris image and contextual eye image to extract features. Then, K-Nearest Neighbor (KNN), Support Vector Machine (SVM) and Kernel based Extreme Learning Machine (KELM) algorithms are adopted for the recognition stage. Experiments are conducted on publicly available data set CASIA-v4 to evaluate the effects of the above features and classifiers. The experimental results suggest that, the contextual eye image features are better than the segmented iris features for human identification and also recommend that feature level fusion is better than the single feature descriptor. The recognition performance of KELM is 98.60% for fusion case of CNN iris and CNN contextual eye image features which is the maximum result in this distance images. The results are also demonstrated that KELM is capable of iris recognition with excellent accuracies for above all feature descriptors.