Biometrics-based verification is an effective approach to personal authentication using biological features extracted from the individual. In this paper, a new personal recognition method is proposed based on the finger crease pattern because the finger creases are thought to be stable and different for each individual. The proposed approach consists of three stages: pre-processing, the feature extraction and feature matching. In the pre-processing stage, a located and normalized method based on the central axes is proposed, and the Region of Interest (ROI) for each finger image is obtained. The region of interest is a rectangular window on the inner side of the finger, which includes the first joint lines and second joint lines in the finger image. In the second stage, Radon transform and the singular value decomposition is applied to extract finger crease features from the region of interest to form the feature vector. The proposed method has been tested on an image database which has 488 images from 61 persons using the nearest neighbor classifier in third stage. The experimental results demonstrate that our proposed approach is feasible and effective in biometrics in terms of the equal error rate(EER = 2.51 percent).