| 一种新的二步指纹分类方案设计 |
| |
| 引用本文： | 肖卫民,袁华强.一种新的二步指纹分类方案设计[J].东莞理工学院学报,2009,16(3):23-27. |
| |
| 作者姓名： | 肖卫民 袁华强 |
| |
| 作者单位： | 华南理工大学,计算机科学与工程学院,广州,510640;东莞理工学院,工程技术研究院,广东东莞,523808 |
| |
| 基金项目： | 国家自然科学基金，广东省科技计划，东莞市科技计划，粤港关键领域重点突破项目东莞专项 |
| |
| 摘 要： | 设计了一种新的指纹二步分类方案，该方案先用基于MKL方法的分类器为待分类指纹确定两个最具可能性的类，再利用经过训练的SPD分类器对第一步的结果进行二分类．用本文的分类方法对标准指纹分类库NISTDB4上的2000枚指纹进行实验得出五分类正确率为95．5％，四分类正确率为96．9％，分类结果优于目前参考的各种分类技术．
|
| 关 键 词： | 指纹分类 方向场向量 子空间分类器 特征选择 |
A New Two-Step Fingerprint Classification Scheme Design |
| |
| XIAO Wei-min,YUAN Hua-qiang.A New Two-Step Fingerprint Classification Scheme Design[J].Journal of Dongguan Institute of Technology,2009,16(3):23-27. |
| |
| Authors: | XIAO Wei-min YUAN Hua-qiang |
| |
| Institution: | XIAO Wei-min~(1,2) YUAN Hua-qiang~(1,2) (1.School of Computer Science , Engineering,South China University of Technology,Guangzhou 510640,China,2.Engineering & Technology Institute,Dongguan University of Technology,Dongguan 523808,China) |
| |
| Abstract: | We describe a new two-step fingerprint classification scheme in this paper.In the first step we adopt a MKL-based classifier to select the two-most-likely classes,which means that it divides the five-class problem to a two-class problem,followed by utilizing a specifically trained SPD classifier to do the final classification.The experimentation is performed on NIST DB4,which is one of the most important benchmarks in this area,showing that the new approach gets better results than the methods previously pu... |
| |
| Keywords: | fingerprint classification orientation vector subspace classifier feature selection |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
