摘要
提出了一种利用隐马尔可夫模型 (HMM )和支持向量机 (SVM )的两级指纹分类新方法 .该方法采用指纹编码 (FingerCode)作为指纹的特征表述 ,在对分类器进行训练之后 ,首先用 5个伪二维HMM对待分类指纹进行类别初选 ,确定最可能的两种指纹分类结果 ,再用相应的SVM分类器做最终判决 .最后使用NIST 4数据库中的 2 0 0 0幅指纹和CQU VERIDICOM数据库的10 0 0幅指纹对该方法进行了实验 ,其分类的准确性为 91% ,连续性为 93.7% 。
A new two-stage method of fingerprint classification is proposed that is based on hidden Markov model (HMM) and support vector machine (SVM). This technique uses FingerCode as the representation of the fingerprint. After classifiers are trained, five pseudo 2-D HMM classifiers are used to firstly select the most possible two classification results. Furthermore, the corresponding SVM classifier is selected to make the final decision. In the end, this new approach is tested by 2000 images selected from the NIST-4 database and 1000 images from the CQU-VERIDICOM database. A classification accuracy of 91 percent and a classification consistency of 93.7 percent are achieved. The results demonstrate the effectiveness of this approach.
出处
《自动化学报》
EI
CSCD
北大核心
2003年第6期851-858,共8页
Acta Automatica Sinica
基金
Supportedbythe 2 11KeyProjectofChongqingUniversity
