Feature And Score Level Fusion Of Multibiometrics

Biometrics is becoming more and more important in human society.With the widespread applications of biometrics,it is becoming increasingly apparent that a sin-gle biometric trait is not sufficient to meet a number of system requirements.Multi-biometric systems seek to alleviate some of the drawbacks encountered by unibiomet-ric systems by consolidating the evidence presented by multiple biometric sources.In recent years,multibiometrics has achieved rapid development,however there are still some key issues to be solved,such as the extraction of complementary information from multiple traits,alleviating the influence of noise data,and the full utilization of infor-mation.To address the issues mentioned above,we propose some new fusion methods based on the data in the feature level and the score level,and achieve high-performance and user-friendly multibiometric systems.The main contributions are summarized as follows:·This thesis proposes a novel machine learning solution to the problem of fusing multibiometrics,aiming at extracting complementary information from multi-modal biometrics.The success of a multibiometrics method depends critically on its ability to fuse complementary information supplied by different modali-ties.The proposed method can capture complementary information,and reduce the dimension of heterogeneous features,with the low consumption of compu-tation ans storage.It also exhibits excellent sparsity and generalization.Due to the extraction of complementary information and the removal of redundant infor-mation,the proposed fusion method can achieve great recognition performance,with the low time consumption.·Identity authentication at a distance has become a trend of multibiometrics.One of the key problems of recognition at a distance is the influence of noise data,which becomes more and more severely after the procedures of feature extrac-tion,template matching and so on.In this thesis,we propose a new robust linear programming modal to fuse multibiometrics by combining the modalities opti-mally,which alleviates the influence of noise data.The proposed method can provide a remarkable tradeoff between conservatism and robustness.The method of solving the proposed modal is also introduced.The proposed fusion modal is also introduced from the view point of statistics,along with the upper bound of misclassification,·Accuracy and usability are the two most important issues for a multibiometric sys-tem.Most of multibiometric systems are based on matching scores or features of multiple biometric traits.However,plenty of identity information is lost in the procedure of extracting scores or features from captured multimodal biometric data,and the loss of information stops accuracy and usability of the multibio-metric system from reaching a higher level.It is believed that there is still some identity information hidden in matching scores,and the information is not fully utilized in the fusion procedure.This thesis proposes a framework of bin-based classifier method for the fusion of multibiometrics,to deal with this problem.The proposed method embeds matching scores into a higher-dimensional space by the bin-based classifier,and rich identity information,which is hidden in matching scores,is recovered in this new space.The recovered information is sufficient to distinguish impostors from genuine users correctly.Therefore,the multibiomet-ric systems which are based on the rich information,are able to achieve a more accurate and reliable result.The ensemble learning method is then used to select the most powerful embedding spaces.It is possible to capture both iris and face biometrics at a distance simultaneously in a single image.So it is expected to achieve a more accurate,secure and easy-to-use multi-modal biometric system if iris and face biometrics can be well combined.Therefore,most of multibiometrics systems are based on iris and face traits,and we also test our proposed fusion methods based on the two traits.In a word,this thesis systematically and deeply addresses the key issues in the fusion of multibiometrics,and our research has promoted the further development of the fusion of biometrics.