摘要

The paper presents a mobile phone palmprint verification system based on novel image capture and feature extraction approaches. The system captures part regions of palm with four fingers closing together and parallel to screen. Then ROI image can be located by detecting key points of fingers region. For feature extraction step, optimal Gabor features of different scales and orientations are extracted from ROI images, and then SDLPP is applied for the dimensionality reduction of Gabor feature space to provide a more compact and discriminant representation. The system achieves the verification accuracy of EER = 3.43% and takes less than 1s to verify a palmprint image, which can satisfy the requirement of a real-time verification system.

  • 出版日期2014

全文