Personal authentication using hand vein and knuckle shape point cloud matching

This paper presents a new approach to authenticate individuals using hand vein point cloud matching with the help of a hardware system. The 3D point clouds of hand veins and knuckle shape were obtained by a binocular stereoscopic vision device. Edges of the hand veins and knuckle shape are used as key-points instead of other feature descriptors because they are better representing the spatial structure of hand vein patterns and significantly increasing the amount of key-points. A kernel correlation analysis approach, which has high sensitivity in hand vein spatial point clouds recognition and doesn't need pre-training the classifier, is proposed to register the 3D point clouds. A hand vein point clouds dataset contains 18 persons' information has been established. The experimental authentication result based on the dataset shows that the method works well and the algorithm is less time-consuming compared with 2D image matching strategies.