Fingerprint Registration Based On Orientation Field And Its Application In Low Quality Fingerprint Matching

As an identity and authentication technology which has comprehensive performance and is used widely,fingerprint recognition has brought convenience and security to the informative life.Fingerprint recognition has been studied for a period of time,and the performance of the high-quality fingerprint matching has been able to meet people's needs,but the performance of the low-quality fingerprint matching is not ideal.It is still a difficult and hot issue to improve the performance of the low-quality fingerprint recognition,The traditional fingerprint matching algorithm has been degraded in low quality fingerprint matching,mainly due to the large noise caused by the loss of true detail features and the generation of pseudo-detail features.As one of the most important fingerprint features,the directional field information is more anti-noise than the detail features and has better robustness to the noise produced by the low quality.In this paper,the traditional global direction field registration method is optimized.The similarity of the global direction field and the classical fingerprint matching algorithm is combined to reduce the EER(equal error rate)of low quality fingerprint recognition,improving the performance of the low-quality fingerprint matching.The main works and innovation of this paper are as follows:1.A method is proposed to register through the steepest descent method based on the center singularity point and the focal point positioning datum point,which effectively solves the problem that the steepest gradient descent method is easy to fall into the local optimum.2.The iterative process of the descending method is optimized.The amount of rotation of the registration is automatically calculated by the descending method,and the iterative divergence problem caused by the block direction is eliminated.3.The fusion of the global direction field similarity score and the fraction of classical fingerprint matching algorithm greatly reduces the EER of low quality fingerprint,and effectively improves the recognition accuracy of low-quality fingerprint.4.The validity of the algorithm is verified by the experiment and matching experiment.In order to verify the performance of the fingerprint direction field registration in this algorithm,the method is compared with the classical registration method,and the FVC2002 database DB1 has the fingerprint direction registration test.The correct rate of registration is 99.47%,better than 98.9% of the classic method.In order to verify the promotion of low-quality fingerprint matching with the fusion matching proposed in this paper,the fusion matching experiment is tested in the internationally recognized low-quality fingerprint database FVC2004 database DB1 and DB2.The matching scores of the Verifinger commercial algorithm with excellent performance are matched with the directional field similarity scores obtained by the method.It is found that the fusion result can effectively reduce the EER.The EER of the Verifinger commercial algorithm in DB1 and DB2 were 4.2% and 3.54% respectively,but after the fusion matching,the EER were reduced to 2.58% and 2.37% respectively.The results show that the steepest decent method based on double-locating is used to register the orientation field,and more accurate registration result can be obtained.On this basis,the global orientation field similarity is calculated,and the global orientation field feature is integrated into the low quality fingerprint matching algorithm,which can effectively improve the recognition precision of low quality fingerprint.The progress of low-quality fingerprint recognition technology can guarantee the accuracy and speed of fingerprint recognition in the case of poor collection performance and low acquisition ability,and ensure the feasibility of fingerprint identification technology in poor image quality,which can promote the promotion of fingerprint identification technology on a larger scale.Compared with the traditional identity authentication technology,it not only reduces the cost of identity authentication,but also improves the efficiency of identity authentication.