An efficient biometric identification technology for Iris scanning.

In this modem age, the use of biometrics is an important aspect in the lives of people in all areas. Iris recognition is now a trustworthy method of personal recognition due to the outstanding requirements for security and reliable types of identification in biometric systems. It is a main and challenging field of study because of its advantages of high precision and non-contact recognition technology. In automatic identification, iris recognition technology plays a vital role. This paper introduces a new concept of optimal security where the proposed system is trained using Convolutional Neural Networks (CNN) incorporated with softmax classifier. The model is trained on a dataset for iris recognition where only a small number of images were trained for each set and showed an enhancement in the results over earlier methods. It focuses on a systematic approach for iris recognition, even when the images have noise, obstructions and various levels of illuminations. Experiments with standard datasets have shown that the put forward approach can attain high performance than competing with other deep learning-based methods. The efficiency gained from iris recognition based on performance evaluation is accurate, achieving a total of 99%. [ABSTRACT FROM AUTHOR]