A Novel Method of Minutiae Filtering Based on Line Feature Extraction

Offline palmprint recognition is one of the most reliable biometric applications in criminal investigation field. Most of current systems use minutiae feature and high-level structure feature to recognize and identify offline palmprints. This paper proposes a novel method to extract lines and wrinkles feature to improve the accuracy and stability of offline palmprint recognition system. Four discriminative criterions are used to define the feature, and thinning image of palmprint in low-resolution is used to extract the feature. The method can be applied for fake minutiae filter, which is important to improve the accuracy of offline palmprint matching, and can also be used to build up a powerful palmprint recognition system based on multi-types of feature. The experiment result proofs the usability and efficiency of the method.