Sub-iris Technique for Non-ideal Iris Recognition

Iris recognition has become one of the most prominent biometrics due to its high recognition accuracy, and it is suitable for the verification and identification systems. Most iris recognition systems can recognize human identity especially when the iris region is free from noise. However, the iris region is typically occluded by noise, eyelid and eyelash. The occluded iris can hide rich iris features, which consists fewer iris features in the iris region, thus reducing the recognition accuracy of the system. The occluded iris is also called a non-ideal iris. In this study, an improvement is made to enhance the performance of an existing sub-iris technique. This can overcome the problem of obtaining iris features from the non-ideal iris. Moreover, the proposed method can also reduce the execution time of the iris recognition system. According to the results, the proposed method managed to localize the sub-iris region, while achieving an equal error rate of 3.75%; area under curve of 0.9832; execution time of 0.719 s; and decidability of 2.9332. These metrics are superior to those of other methods.