Structure Feature Extraction For Finger-vein Recognition

As a new biometrical identification technology, finger-vein recognition has drawn increasing attention from biometrics community in recent years. The finger-veins used for biometrics are superficial veins beneath the skin layer. Compared with the traditional biometric characteristics(e.g. face, iris, and fingerprints), finger-vein patterns exhibit some excellent advantages in real application. For instance, finger-vein patterns are universality, uniqueness, permanence and measurability, finger-vein images can be captured noninvasively without the contagion and unpleasant sensations. These merits make the finger-vein trait highly unique and fraud-proof for personal identification.The finger-vein uniqueness is directly determined by the randomness of the finger-vein networks. In anatomy, finger veins form a network along a finger and the network cannot be broken unless some veins suffer rupture. Hence, network based finger-vein description is crucial for the understanding of finger-vein characteristics. The structure features related to the finger-vein networks are therefore very crucial in finger-vein pattern representation. In graphic interpretation, a network can be further decomposed into curve segments and junctions. Thus, the structure features related to the curve segments and junctions are capable of describing the vein network pattern in terms of its topology and geometry.In this thesis, a new finger-vein recognition method based on structure feature is proposed. First, the reliable finger-vein skeleton is extracted from the preprocessed images. Then, the minutiae points including junction points and ending points are extracted and three curve segments around each junction point are traced. These curve segments are encoded with MIAC to form the complete structure feature. A dynamic matching method is adopted at last. Experimental results show that the proposed method is highly powerful in improving finger-vein matching accuracy.