Real-time Iris Recognition System for Non-Ideal Iris Images
Noel B. Linsangan, Ayra G. Panganiban, Paulo R. Flores, Hazel Ann T. Poligratis, Angelo S. Victa, Jumelyn L. Torres, Jocelyn F. Villaverde · 2019
Biometrics is popular nowadays because of its very useful security applications. There are different biometric technologies but iris recognition system was considered the most reliable since human irises are unique and cannot be forged easily. The study aims to segment ideal and non-ideal iris images with the help of Zuo and Xin Li's algorithm to determine the most accurate wavelet family and its coefficient for encoding the iris templates using Haar, Daubechies, Biorthogonal and Reverse- Biorthogonal wavelets. Test metrics, like False Rejection Rate (FRR), False Acceptance Rate (FAR), Compression Rate (CR), and Degrees-of-Freedom (DOF) were used in evaluating the performance of the system. Based on the results, the algorithm was able to segment ideal and non-ideal iris images, encode the irises and match the irises accurately. Haar was proven as the most accurate iris recognition algorithm while Reverse-Biorthogonal was the better wavelet algorithm image compression. There were different wavelets that would give better result in the recognition process depending on the algorithm or system used. The metrics suggested that the developed algorithm was sufficient for iris recognition.