Abstract—In this paper, a novel method is presented using deep neural network to identify singular points on a fingerprint. The proposed method can efficiently calculate fingerprint blocks orientation using the pre-trained neural network. The same neural network is applied again to define singular points at pixel level. The training step may be complicated and time consuming, but adopting a pre-trained model to calculate orientations outperforms algorithms that calculate pixel orientation in real time. In addition, the proposed model is rotation insensitive, and experiment results show that the proposed method is so robust that it can identify singular points as small as a circle with few pixels in radius.
Index Terms—Fingerprint image processing, singular point, machine learning, deep neural network.
L. Liu is with the Department of Information Management, Shih Hsin University, Taipei, Taiwan (e-mail: liminliu@mail.shu.edu.tw).
Cite: Limin Liu, "Fingerprint Analysis and Singular Point Definition by Deep Neural Network," International Journal of Machine Learning and Computing vol. 8, no. 6, pp. 625-629, 2018.