A Reference Point Detection Algorithm for Top-View Finger Image Recognition

This paper describes an algorithm for automatic reference point detection in a top-view finger image recognition system. In tests of 700 finger images, only 6 images were rejected by our algorithm. A reference point location error correction technique was developed to improve the recognition accuracy. When using the proposed algorithm, the accuracy of the top-view finger image identification system was only reduced to 93.80% compared to 96.57% when using a manually defined reference point. This shows the feasibility of using top-view finger images to increase the recognition accuracy of fingerprint identification.