Pengenalan iris mata dengan algoritme voting feature interval versi 5 menggunakan ekstraksi ciri log-gabor wavelet

Abstract

Biometric recognition based on iris patterns has its own advantages because iris patterns are more stable and reliable as compared to other biometric subjects such as face and fingerprint. This research provides implementation for recognizing eye images taken from CASIA dataset based on Daugman methods for extracting features. The system uses an automatic segmentation based on threshold to localize the iris collarette and normalize the results to constant dimension using Daugman’s rubber sheet model by remaping each point within the iris region to a pair of polar coordinates. The features are extracted using 1D log-Gabor wavelets to create template which contains dimension 2 times from its normalized images. The template data are then splitted into three subsets and then alternately used for training and testing using voting feature interval version 5. The best recognition of testing data are obtained from combining vote from left and right eyes rather than using single eye sides.