Fast Fingerprint Minutiae Matching Based on Tent Map Chaotic Particle Swarm Optimization

Fingerprint matching is one of the key parts in fingerprint identification system. To further improve the computational efficiency and matching accuracy of the fingerprint matching algorithm, a fast fingerprint minutiae matching algorithm based on improved Tent map chaotic particle swarm algorithm is proposed in this paper. Firstly, the particle swarm optimization is introduced into point pattern matching based on fingerprint minutiae. The chaotic genus-randomness and ergodicity are used to overcome the defects of basic particle swarm algorithm, which it is easy to fall into local extremum, slow to converge in later stage and its precision is low. In view that Tent map has better ergodicity than Logistic map and chaotic optimization based on Tent map can further improve searching efficiency, parameter estimation of the geometric transformation for fingerprint minutiae matching is optimized by the modified Tent map chaotic particle swarm algorithm, in order to improve the convergence accuracy and operation speed. Then, a hierarchical matching method is adopted. The corresponding fitness function for minutiae matching is designed and implementation procedures of the algorithm are given. The matching relation of the minutiae pair is determined by the local structural information of the translation-rotation-invariant minutiae after the coarse matching, to resist the effect of rotation, translation, local non-linear deformation and noise of the low- quality fingerprint image introduced in the course of image acquisition. Finally, a large number of experimental results of fingerprint minutiae matching aiming at the FVC2006 fingerprint database are given. The corresponding objective quantitative evaluation is performed. The results show that the proposed algorithm not only obtains higher matching precision, but also its computation speed is increased by twice, when compared with the fingerprint minutiae matching algorithm based on genetic algorithm proposed recently. The proposed algorithm meets the real-time requirements and has been successfully applied to fingerprint personal identification system.