Personalized Fusion Method Based on Finger Vein and Finger Contour

Finger vein is a promising biometric for the authentication due to its some advantages. However, when capturing the finger vein image, the pose variation of the finger or the illumination variation may cause failure to verification. To overcome these limitations of using a single biometric, multiple biometrics can be combined to improve the performance. Compared with the fusion of other biometrics, the advantage of the fusion of finger vein and finger contour is that acquiring the two biometric image is convenient because the finger vein image and finger contour image can be obtained by a finger vein capturing device. In this paper, we propose a personalized fusion method based on finger vein and finger contour at the score level. The samples are firstly classified based on the original matching score, and then the classification confidence score (CCS) can be obtained based on the classification result. Compared with the traditional score, CCS contains additional classification information which may provide more useful information for the final fusion. In addition, a more effective personalized weight fusion based on CCS is proposed due to the difference of different subjects. Finally, we conduct extensive experiments on our database to evaluate the effectiveness of our proposal.