Research On Palm Multi-modal Recognition And Anti-spoofing Based On Dual Wavelength

In recent years,biometrics have become more and more widely used in the field of personal identification,and the public's requirements for the security of identification systems continue to increase.However,with the emergence of artificial fingerprint and artificial masks,the security of system is seriously threatened.The hand contains a variety of biological characteristics and is easy to use.At present,the palm prints and palm veins are widely used in many occasions,but there are still cases of spoofing.People began to develop biometric identification systems with higher anti-spoofing capabilities.Through research,it was found that the process of acquiring palm prints and palm veins can be synchronized with the process of acquiring spo2 and pulse.In addition,spo2 and pulse have higher security capabilities.Therefore,this thesis proposes a synchronous acquisition system.The system can acquire both static and dynamic biometrics without additional hardware requirements.The main difference is requiring a certain period of time for dynamic measurement.The main work and results of this thesis are summarized as follows:(1)Research on palm multi-modal synchronous acquisition and processing.Acquiring biometrics with higher anti-spoofing capabilities and improving system security are key point.The palm contains a variety of biometrics,including palm prints,palm veins,spo2,and pulse biometrics.Because the biological characteristics of spo2 and pulse are inherent characteristics of the human body,so they have high anti-spoofing capabilities.Therefore,a multi-feature acquisition system for hand is designed in this thesis.The acquisition system is a simultaneous acquisition of four biological characteristics,including palm print,palm vein,spo2,and pulse.(2)A palm multi-modal recognition and anti-spoofing system was established.This system mainly includes acquisition system and parallel processing program.The construction of the acquisition system should not only consider the placement of the optical platform and hardware control circuits,but also the acquisition method between biological characteristics.Acquisition of palm prints requires visible light,and acquisition of palm veins requires near-infrared light sources.The acquisition of spo2 and pulse requires the PPG signals.PPG signals require two light sources with different wavelengths.Therefore,a synchronous acquisition method based on dual wavelength is designed in this thesis.A dual wavelength is a set of visible light sources and a set of near-infrared light sources.The parallel processing program includes an image preprocessing part and a palm modal recognition part.The purpose of parallel processing is to use GPU devices to speed up the processing of data by the system.The palm modal recognition part is to prepare for identification and anti-spoofing.(3)Evaluation of biometric anti-spoofing capabilities.Comparison of anti-spoofing capabilities of palm print and palm vein static biological characteristics with spo2 and pulse dynamic biological characteristics.Based on analysis of mechanical,optical,and electrical properties of the human skin,various artificial palm print and palm vein are fabricated with a variety of materials in this thesis.The artifical palm prints and palm veins are used to deceive the real-identification instruments and classification networks.Through experiments,it is proved that the anti-spoofing ability of palm prints and palm veins is insufficient.However,the system can simultaneously acquire dynamic biological characteristics,and use the dynamic biological characteristics to improve the anti-spoofing ability.