Dorsal Hand Veins Recognition Based on LogGabor Feature Sparse Representation and Classification

Dorsal hand vein recognition is a biometric technology emerged in recent years. In this paper, we proposed to employ the LogGabor filter to extract multi-scale and multi-orientation information from near-infrared dorsal hand vein images. A multi-channel feature descriptor was constructed based on the sparse representation and dictionary learning. Then, a multitask sparse representation classifier was employed to classify and recognize the dorsal hand vein patterns. The proposed method was validated on the self-established dataset and two public datasets. The experimental results demonstrated that the proposed method can describe effectively the texture information, and recognize the patterns of dorsal hand veins with high accuracy. In addition, the computational efficiency can be improved effectively by adopting a fast filtering algorithm.