摘要
掌纹识别作为一种新兴的生物识别技术近年来得到了广泛研究.鉴于毯子维(Blanket dimension,BD)的多分辨率特性和掌纹纹理的方向性,本文对比分析了普通毯子维、水平方向扩张毯子维和垂直方向扩张毯子维对掌纹分形特征的表达性能,提出并研究了一种基于水平扩张毯子维的掌纹识别新算法.本文算法在香港理工大学掌纹数据库(版本2)进行了实验,实验结果表明,水平扩张毯子维可以很好地提取掌纹特征,获得的识别率可达99.9%,识别时间小于287ms,可满足在线掌纹识别.
As an emerging biometric technology, palmprint recognition has been extensively researched. Due to mul- tiresolution characteristics of the blanket dimension （BD） and the directionality of palmprint texture, feature extraction efficiencies with blanket dimension, horizontally and vertically expanded blanket dimensions （HEBD and VEBD） are com- pared, and a palmprint recognition algorithm based on horizontally expanded blanket dimension is proposed according to the comparison results. The algorithm is evaluated on Hong Kong PolyU Palmprint Database V2.0. Experimental results show that the horizontally expanded blanket dimension can extract the features that well represent the palmprint, and a recognition rate of 99.9% is achieved. The total execution time of recognition is less than 287 ms, which is fast enough for on-line palmprint recognition application.
出处
《自动化学报》
EI
CSCD
北大核心
2012年第9期1496-1502,共7页
Acta Automatica Sinica
基金
国家自然科学基金(30870666)
山东省科技发展计划项目(2010GSF10243)资助~~
关键词
分形维数
毯子维
关掌纹识别
特征提取
Fractal dimension, blanket dimension （BD）, palmprint recognition, feature extraction
