Transforming traditional iris recognition systems to work on non-ideal situations

Non-ideal iris images can significantly affect the accuracy of iris recognition systems for two reasons: 1) they cannot be properly preprocessed by the system; and/or 2) they have poor image quality. However, many traditional iris recognition systems have been deployed in law enforcement, military, or many other important locations. It will be expensive to replace all these systems. It will be desirable if the traditional systems can be transformed to perform in non-ideal situations without an expensive update. In this paper, we propose a method that can help traditional iris recognition systems to work on the non-ideal situation using a video image approach. The proposed method will quickly identify and eliminate the bad quality images from iris videos for further processing. The segmentation accuracy is critical in recognition and would be challenging for traditional systems. The segmentation evaluation is designed to evaluate if the segmentation is valid. The information distance based quality measure is used to evaluate if the image has enough quality for recognition. The segmentation evaluation score and quality score are combined to predict the recognition performance. The research results show that the proposed methods can work effectively and objectively. The combination of segmentation and quality scores is highly correlated with the recognition accuracy and can be used to improve the performance of iris recognition systems in a non-ideal situation. The deployment of such a system would not cost much since the core parts of the traditional systems are not changed and we only need to add software modules. It will be very practical to transform the traditional system using the proposed method.