An Biometric Model for Iris Images Segmentation and Deep Learning Classification

Iris biometrics is a rapidly developing technology that many people are interested in. Using iris biometrics for person identification does not require touching a human body. In practical applications, unexpected oscillations in iris images pose a challenge for automatic iris identification. Existing methods that deliver the eye image to a deep learning network reduce accuracy and provide inaccurate iris data. It's hardest to use unrestricted iris recognition systems because they produce a lot of noise. Other problems include changing lighting, eyelids or eyelashes covering the iris, specular highpoints on the pupils from a light source during image capture, as well as the subject's gaze moving around while the image is being taken. The iris identification process heavily relies on iris segmentation. There are a lot of different kinds of noise in an eye image, so the segmentation can turn out wrong. This work does the initial work on the outer border segmentation of the iris. The next step involves locating the pupil's boundary. The segmented image is fed to the Alexnet system, grounded on deep learning neural networks, for categorization. Using the supplied eye images, the system first takes a picture of the pupil's center and border. To confirm identification, next compare the iris's center and border with those of the previously established reference pupil. Results from the experiments show that the proposed model is better than the previous ones.