Research On Template Protection And Presentation Attack Detection For Finger Vein

Finger vein recognition has gradually become a research hotspot in recent years due to its significant advantages.However,most of the current finger vein recognition systems directly store the original templates.Once an attacker attacks the database of the system through certain illegal operations,the loss of the templates will cause serious privacy violations for users due to the particularity of biometrics.In addition,the lost templates are likely to be used by attackers as a tool for presentation attack,causing serious system security risks.In order to solve these security problems,this thesis studies the finger vein template protection and presentation attack detection algorithms,and does the following work.(1)The current finger vein template protection algorithms are difficult to meet the characteristics of the template in terms of security performance,recognition performance and diversity at the same time.In order to balance these needs,a finger vein template protection algorithm based on chaotic map is proposed.This algorithm takes advantage of the similarity between chaotic map and cryptography,and uses its parameters as the system key.The key space is large.Finally,the templates are identified in the encrypted domain.Compared with the existing algorithms,the simulation results show that the security performance of the algorithm in this thesis is improved,and the recognition performance of the system after template protection is stable.Besides,it can also meet the diversity requirement of the template in the case of key loss.(2)Aiming at the low performance of finger vein presentation detection using a single texture feature,this thesis proposes a finger vein presentation detection algorithm based on multi-feature fusion.The algorithm uses a weighted fusion method to fuse noise features and fuzzy features in finger vein images,and uses the particle swarm algorithm to optimize the weights to improve the performance of presentation detection.At the same time,in order to enrich the experimental simulation environment,in addition to the public authenticity finger vein database,a new database is built for algorithm evalution.The simulation results show that the performance of the proposed method is improved after multi-feature fusion.It can achieve effective detection in both public and self-built databases.Compared with the existing algorithms,it has significant advantages in detection performance.