Finger pattern identification for authentication purpose

The pattern on the finger face is composed of various size of creases, knuckle lines and subcutaneous vascular tubes. The patterns, shapes and sizes of fingers are different from person to person. These features are unique and peculiar to each person such that it can be used as the information for the process to identity an individual. In this paper, the properties of the features on finger face are analyzed and a method of distinguishing a person by the features of finger patterns, shapes, and sizes is presented. All of these features of a finger are able to be represented by a set of matrices obtained through a computation using FFG(Facet Function Gradient) masks. The gradient and the curvature at each patch in an image are able to be computed by certain operations with the elements of the matrices. In this paper we introduce an algorithm to identify the finger features using the matrices. It is possible to recognize not only the finger pattern but also the shape and size of the finger by this method. Recognition rate is turned out to be 98.2% in an experiment using the method suggested.