CARACTERIZACIÓN Y EXTRACCIÓN DE CARACTERES DEL IRIS PARA APLICACIONES DE RECONOCIMIENTO

Abstract

Nowadays recognition systems of people have became common methods of identification. Every day we see how its use has spread to the commercial, private, government, etc. Therefore the main objective is the study of digital image processing method of iris recognition as individuals. The Iris has an interesting structure from the point of view of their morphological constitution, textures with that account, have information that allows us to identify a person; therefore it is desirable to develop an effective method to identify individuals using biometric traits. This work aboard many methods of image processing based on the iris detection and segmentation. For the iris localization first we filter the image in order to enhance and get better quality in the iris textures. The next step is the border detection that is realized with the canny filter, then we applied a dilatation operator, this operator is used to locate parametric forms based on a point distribution present in a binary image (lines, curves, ellipses). Finally the combination of these methods allows us to realize quickly and precisely the iris detection and segmentation. When we get the segmentation image, we apply the Gabor filter in order to featuring textures taking as the principal hypothesis that the iris could be classified as a texture and using it to generate an eigenvector as input to an Support Vector Machine (MSV) or RNA-Retro-propagation. The results show that the methodology is working properly and yielded an 88% rate of recognition for the samples used in the CASIA-IrisV3-Lamp database.