An efficient finger-vein extraction algorithm based on random forest regression with efficient local binary patterns
Chenguang Liu, Yeong‐Hwa Kim · 2016
Finger-vein, as a secure and convenient biometric characteristic in nature, has been widely studied for authentication in recent years. In this paper, we propose an efficient finger-vein extraction algorithm based on random forest training and regression with efficient local binary pattern feature. By integrating with a vein pattern matching method which is robust to finger misalignment, we achieved state-of-the-art finger-vein recognition. Thorough experiments have been conducted on two popular databases to prove the effectiveness and robustness of the proposed method.