In order to maximize the robustness of fingerprint identification and to minimize the identification time, it is better to align the fingerprint first by referring the singular points of fingerprint. In this paper, we proposed a new algorithm to localize the singular points of fingerprint. The proposed approach is less computational complexity and higher precision. This paper shows how, from the x-gradient component of orientation filed and its sine component, very accurate extraction of singular points can be obtained. This localization can be used for accurate registration of two fingerprints.