Search for: All records

While deep learning techniques are increasingly becoming a tool of choice for iris segmentation, yet there is no comprehensive recognition framework dedicated for off-angle iris recognition using such modules. In this work, we investigate the effect of different gaze-angles on the CNN based off-angle iris segmentations, and their recognition performance, introducing an improvement scheme to compensate for some segmentation degradations caused by the off-angle distortions. Also, we propose an off-angle parameterization algorithm to re-project the off-angle images back to frontal view. Taking benefit of these, we further investigate if: (i) improving the segmentation outputs and/or correcting the iris images before or after the segmentation, can compensate for off-angle distortions, or (ii) the generalization capability of the network can be improved, by training it on iris images of different gaze-angles. In each experimental step, segmentation accuracy and the recognition performance are evaluated, and the results are analyzed and compared. 

Emerging standoff iris recognition systems operate under unconstrained conditions and the iris images captured by these systems are more subject to off-angle acquisition distortions. While deep learning techniques (e.g. convolutional neural networks (CNNs)) are increasingly becoming a tool of choice for iris segmentation tasks, yet there is a significant lack of information about how these distortions affect the performance of such networks. In this work, we thoroughly discuss the general effect of different gaze angles on ocular biometrics and relate the findings to off-angle iris segmentation using CNNs. In particular, we conduct systematical analysis on the impact of different gaze angles on segmentation performance of two CNNs with different architectures. The networks’ performance turns out to have a direct relation to the closeness of gaze-angles in the training and testing images, and it declines as the gaze angles diverge. We further investigate the effect of (i) increasing the quantity of iris training data in case of gaze angles in training and test data match, and (ii) considering iris training data consisting of several distinct gaze-angles (we obtain promising results using the second configuration). Finally, we compare our results to those of some classical iris segmentation algorithms, where the CNNs are found to outperform the classical algorithms.