Pengenalan Iris Mata Menggunakan Probabilistic Neural Network dengan Ekstraksi Ciri Log-Gabor Filter

Abstract

One of the objects of research that is currently prevalent in the developed world is the individual identification technology based on biometrics. Iris biometric is promising for identification, because it has a consistent pattern compared with other types of biometrics. This study established a system of identification using iris images from CASIA eye using 1D log- Gabor filter for feature extraction and PNN as classifier. This system uses automatic segmentation based on the threshold to localize the iris collarette region and normalize the results to the constant dimension using Daugman’s dimensional rubber sheet model by mapping each point in the iris region to a pair of polar coordinates. Data were divided into three subsets. Tests were conducted on the left eye dataset, right eye dataset and a combined dataset of left eye and right eye. The average accuracy produced for the left eye dataset is 98%, for the right eye dataset is 97%, and for the combination of left and right eye is 100%.