Finger-knuckle-print recognition based on image sets and convex optimization

In order to enhance the stability and security of biometric features recognition, the finger-knuckle-print (FKP) is used in this paper to study high performance recognition problem based on image set. After extracting the image feature by the method of local phase quantization, an image set can transform to a closely related set of points in the affine space. Then the models of the convex hulls are constructed by these point sets. Finally, the FKP recognition was processed in the optimized convex model. Experiments on the publish FKP database show that the proposed algorithm achieves a reliable performance and is suitable for the image data sets.