Iris biometric recognition of latin american eyes in a white light controlled environment

Abstract

Iris biometric recognition has taken a great interest in the last decade. Many private and public facilities need to improve their security access due to the increasing terrorist attacks, the growing number of fraud and identity theft among other felonies. The regular technologies are almost all based on infrared light illumination due to the ease of using this for texture approach recognition. Even Though this technique is suitable in most cases, the highest price of these cameras on the market demands the research of either daylight or a white light solution. There are some known iris databases such as CASIA and UBIRIS, but they are not from Latin-American eyes that have different occlusion and color issues. A white light controlled environment was used to take the set of pictures 500 that let us have a new color database to carry out the experiments based on the EigenIris technique. The best iris and pupil segmentations were obtained either on the red or green color layers of the RGB images at 100% of correct localization. In this research, a sensitivity of 0,82 and specificity of 0,85 was achieved with a modified version of the EigenIris method using the blue color layer.