With the aim to locate singular points (core point and delta point) precisely and to complete fingerprint classification and matching, the Radon transform was introduced firstly to extract the directional filed characteristic of a fingerprint image and to implement the image segmentation. The concept of directional entropy was proposed to describe the distribution of directional filed characteristic and the singular point area search method based on directional entropy was given. Furthermore, a directional density function was established to evaluate effectiveness of singular point location, which can guide the inspecting system to achieve optimal directional entropy threshold. After singular points were detected, the performance of similar algorithms was compared by taking accuracy and efficiency into consideration. Experiments show that the algorithm proposed in this paper is more advantageous, and it not only achieves high inspecting accuracy by 83%, but also has better adaptability and robustness fitted for practical application.