FREESTYLE NON-GUIDED TRUE-ANATOMICAL BIOMETRICS: A FUSION OF HAND GEOMETRY AND TEXTURE FEATURES

This paper presents a freestyle non-guided- anatomical hand biometric system based on fusion of hand geometry and texture features. The system uses a photographic CCD camera or IR camera to capture hand images. The captured data is freestyle where no hand control guides or pegs are used. Captured images are first preprocessed, and then geometric hand features are extracted. Geometric hand features include: finger lengths, areas, circumferences, and widths. Images are then analyzed for the textural patterns on the back of the hands. Texture features of first and second orders statistics are extracted. Geometric and texture features are used to construct a feature vector. Performance statistical parameters are found to exceed 99.8%. This encouraging result suggests fusion of the texture information for the back of the hand together with other