HUMAN EYES IMAGES BASED MULTI-FEATURE FUSION IDENTIFICATION METHOD

摘要

A human eyes images based multi-feature fusion identification method is disclosed, which comprises registration and identification. Wherein, the registration includes: obtaining normalized human eyes images and normalized iris images from given human eyes images of registration, extracting multimode features from human eyes images of users waiting for registration, and storing the obtained multimode features of human eyes images as registration information in a registration database. The identification includes: obtaining normalized human eyes images and normalized iris images from given human eyes images of registration, extracting multimode features from human eyes images of users waiting for registration, comparing the extracted multimode features with multimode features in a database to obtain comparion scores, obtaining fusion scores via score level fusion, and performing human eyes images based multi-feature fusion identification using a classifier. The present invention performs identification by fusion of various feature information in an eyes area of a human face, and a system with high identification precision can be applied to places of application with high security level.