Iris Liveness Detection Based On Light Field Imaging

As people are currently paying more attention to security,traditional identification methods have no longer satisfied people's demand,while biometric technology has received more and more attention.Iris recognition technology is becoming the most secure biometric identification technology at present and has been widely used in various fields such as national defense,education and finance due to its advantages such as uniqueness,stability and high precision.However,current iris recognition devices are threatened by its vulnerability to spoofing attacks due to poor protection of acquisition device.Typical spoofing attacks involve methods such as printing iris patterns on contact lens or papers.Iris liveness detection is the most important means of detecting counterfeit samples and would significantly contribute to current iris recognition systems.Therefore,it is very important to study iris liveness detection technology.Light field imaging is a computational imaging technique that is a major technological innovation compared to traditional optical imaging.The light field represents radiances of light rays along traveling paths,which describes intensities and transport characteristics of each light signal in an open space.The visual contents,such as depth,texture,and illumination are encapsulated in the light field.By using light field camera,light field imaging has become the state-of-art method to capture both intensity and direction information of visual objects and transform them to fourdimensional light field data,providing promising solutions to biometrics.In this thesis,light field photography techniques are introduced into iris liveness detection.The light field photography technology can improve the usability and security of the iris recognition system as it can achieve liveness detection in large depth of field.In this work,by pre-experimental works including designing the data acquisition scheme,environment and process,the light field iris liveness database was firstly established by using light field cameras.Based on the light field data collected in this database,it is realistic to perform later experiments.Secondly,the collected light field iris database is comprehensively evaluated from the aspects of data diversity,image preprocessing and image quality,and the ideas as well as principles of light field iris detection are analyzed at the meantime.Finally,this thesis proposed an iris liveness detection algorithm which takes full advantages of intrinsic characteristics in light field iris imaging.Geometric features as well as the texture features are extracted from captured light field image by using light field refocusing technology.Such features are combined for genuine and fake iris image classification.Experiments were carried out based on the self-collected near-infrared light field database.Experimental results have proved that the method proposed in this work achieved high iris liveness detection accuracy with the image acquisition environment of 30 cm depth of field at 1.6m distance.It is expected that the method proposed in this thesis is able to work effectively and accurately to prevent hybrid spoofing attacks,which effectively expands the application range and scene for iris liveness detection.