中文摘要：

提出一种基于多特征描述的指横纹识别方法.分别提取指横纹的主成分特征、Gabor相位特征和Gabor幅值特征构成识别系统,采用Fisher线性判决方法融合各自匹配分数,进一步提高系统性能.通过98个人、1 971幅图像的测试实验表明,本文方法在获得较高性能的同时（识别率为99.39%,平均错误率为0.56%）,单次匹配时间仅为0.67 ms,可以满足中等规模数据库实时识别要求.

英文摘要：

A novel knuckleprint authentication method is proposed by using multiple representations.Principle component analysis（PCA） features,2D Gabor phase features and magnitude features are extracted for knuckleprint authentication.Fisher criterion based linear discrimination analysis（LDA） is used for match-score fusion,which can further improve the system performance.Experiments based on the database that contains 1 971 image samples from 98 individuals demonstrate that the high recognition accuracy and efficient performance can be achieved with the proposed algorithm.The recognition rate is 99.39%,and the half total error rate（HTER） is no more than 0.56% and one match time consumption reaches 0.67 ms.