基于线性细节特征的掌纹信息表达与识别  

Representation and recognition of palmprints based on line minutia features

作　　者：裘旭光[1] 刘晶[1] 张大鹏[2] 

机构地区：[1]天津工业大学计算机技术与自动化学院,天津300160 [2]香港理工大学计算学系

出　　处：《天津工业大学学报》2006年第6期44-47,共4页Journal of Tiangong University

摘　　要：提出了一种基于掌纹线性细节特征进行身份鉴别的新方法,对掌纹图像进行二值化和骨骼化处理后,从中提取掌纹主线和皱褶的线性细节特征,包括直线段、曲线段和分叉等,通过定义特征向量来表达掌纹信息,在此基础上使用掌纹细节特征向量的匹配算法实现了身份鉴别.采用该方法在香港理工大学的掌纹数据库中进行一对一测试,准确率达到97.500/.A novel approach of identification recognition based on the line minutia features （LMF） of palmprints is proposed. The palmprint image is handled by threshold and skeleton, after that the LMFs of the principal lines and wrinkles, including line segment, curve segment, bifurcation et al, are extracted. The line minutia feature vector （LMFV） is defined to represent palm-lines. The matching method based on the LMFV is employed to complete the verification. During the matching phase, 97.5% accurate rates are obtained in the one-to-one matching test based on the palmprint database of Hong Kong Polytechnic University.

关 键 词：生物特征 掌纹识别技术 线形细节特征 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]