| This research is concerned with multi-feature liveness detection methods using spatial biometric characteristics in a face to enhance the performance of the system against spoofing attacks.Biometric systems for a variety of applications are well-researched and can provide adequate solutions.Presentation attacks(spoofing),however,where an attempt is made intentionally to subvert the device,is a significant challenge to their use in unattended applications.The liveness detection method can help protect biometric systems from such attacks.In this research,the novel techniques for liveness detection are presented using a triplecore(three main algorithms)called focal distance,coordinate position,followed by a challenge response from a user.The first core algorithm is called focal distance,which the purposes are introduced and used to verify the different distances between facial features and the camera since the real face has a different focal distance for each facial feature.Meanwhile,the second core algorithm using coordinate position is aimed at observing the movement of the facial features and to verify the position of the facial features in vector space.Finally,the last algorithm is a challenge response which the purpose of the questions,or challenges,is only for users who successfully answer the question or certain tasks.The proposed method is tested using three public datasets(NUAA,Replay Attack,and CASIA)and one series dataset taken from fellow friends,classmates,lab-mates,and family members named self-collected dataset.Three attack scenarios were explored including picture,mask,and video replay and data were collected to evaluate the system’s performance through various combinations of characteristics and methods.The multi-classification approach,which combined data from different sets using score fusion,provided the best results and showed the state-of-the-art to be well equally effective and feasible.The evaluation results indicate that the proposed method using a triple-core algorithm able to perform the liveness detection,distinguish the spoof,also to reduce the error rate effectively. |
