Fingerprint core point localization and its orientation computation

Core point, as an essential feature of fingerprint, plays an important role in fingerprint matching/classification, where the core point region is distinguished from non-core point region by the machine learning method, and their ridge orientation distributions can be used to form training data. Then, the multi-resolution SVM method is used to gain a training model so as to predict accurately the position of core point by corresponding models. Moreover, the orientation of core point is defined reasonably and a heuristic method is devised to compute it. Experimental results showed that the proposed method can localize the position of core point and compute its orientation with high accuracy and efficiency.