Abstract: Finger vein recognition uses the unique of finger vein patterns to identify individual at high security and convenience. To further improve recognition performance, this paper presents a method that combines both of local and global feature of finger vein image. It extracts the local feature of a user’s finger vein image using Local Binary Pattern（LBP） algorithm and gets matching score with Hamming distance. It uses bi-direction two dimensional Principal Component Analysis（（2D）2PCA） to extract the global feature of the finger vein image and gets matching score with Euclidean distance. It fuses the two matching scores at score level and makes the decision. Experimental results show that the local feature and the global feature of finger vein image have complementarities, and the recognition accuracy is effectively improved.

Key words: finger vein recognition, local feature, global feature, score level fusion

摘要： 手指静脉识别是利用人体手指静脉结构的唯一性实现个体身份认证，具有高度安全和使用便捷等优点。为了进一步提高手指静脉识别系统的性能，提出了一种融合局部特征和全局特征的手指静脉识别方法。应用局部二元模式方法提取手指静脉局部特征，利用海明距离计算匹配得分；应用双向两维主成分分析方法提取手指静脉全局特征，利用欧式距离计算匹配得分；在得分级上融合二者的匹配得分以产生识别结果。实验结果表明，局部特征与全局特征具有较好的互补性，有效地提高了识别精度。

关键词: 手指静脉识别, 局部特征, 全局特征, 得分融合