白雪飞¹, 高楠¹, 张宗华¹, 王向军²
AuthorsHTML:白雪飞¹, 高楠¹, 张宗华¹, 王向军²
AuthorsListE:Bai Xuefei¹, Gao Nan ¹, Zhang Zonghua ¹, Wang Xiangjun²
AuthorsHTMLE:Bai Xuefei¹, Gao Nan ¹, Zhang Zonghua ¹, Wang Xiangjun²
Unit:1. 河北工业大学机械工程学院，天津 300130；2. 天津大学精密仪器与光电子工程学院，天津 300072
Unit_EngLish:1．School of Mechanical Engineering, Hebei University of Technology, Tianjin 300130, China
2．School of Precision Instrument and Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China
Abstract_Chinese:针对现有三维掌纹身份认证技术样本容量小和识别速度慢的问题, 提出一种基于曲面类型(surface type, ST)和主成分分析的三维掌纹识别新方法．该方法利用分块ST图像的直方图获得更加精细的掌纹特征信息; 通过预先降维和主成分分析法相结合, 大幅减少后续运算量以及三维信息量; 最后依据特征主成分分析的最近邻距离对比, 完成身份认证．实验结果表明, 该方法能在35 ms内完成三维掌纹的4 000次匹配, 从而实现大样本条件下特定样本的快速甄别．与已有的三维掌纹识别方法相比, 本文提出的方法不仅能够克服传统方法的小样本局限性, 而且可降低运算量和数据冗余; 在实现大样本身份识别的同时, 兼具更高的识别精度、速度和鲁棒性．
Abstract_English:In order to conquer the limitations of small sample size and low identify speed in the existing three dimensional(3D)palmprint identification technology，a novel method based on surface type(ST)and principal component analysis(PCA)is proposed. The algorithm adopts the histograms of blocked ST as effective feature of the palmprint，reduces the subsequent computation complexity and compresses the 3D information substantially by combining descending dimension in advance and PCA. Nearest neighbour classifier is chosen as the discrimination criterion to identify a person. Experimental results indicate that the proposed method could complete 4 000 times of matching and identify a person within 35 ms in the case of 4 000 samples，which means that the algorithm is especially suitable for large sample database. Compared with the existing methods of 3D palmprint identification，the proposed one could overcome the traditional problem of small sample，reduce the computation complexity，and realize the identification in the case of large sample database with higher accuracy，greater speed and stronger robustness.
Keyword_Chinese:三维掌纹识别; 曲面类型(ST); 分块直方图; 主成分分析法
Keywords_English:three dimensional(3D)palmprint identification; surface type(ST); blocked histograms; principal component analysis(PCA)

PDF全文下载地址:http://xbzrb.tju.edu.cn/#/digest?ArticleID=6042