摘要
本文提出了一个基于多分类器组合的手写体数字识别方法．文中首先给出了一个客观评价分类器性能的参数，其后基于此参数提出了多分类器的组合方法，并从理论上研究了此方法的一些性质．本文实验采用Concordia大学模式识别与机器智能中心的手写体数字数据库，在实验中，使用了9个利用不同特征分类器进行组合．组合后识别率、拒识率和可靠性分别可达到97．05％，2．05％，99．08％．
A handwritten digit recognition method based on multi-classifier combination is described in the paper. First I an objective parameter called performance parameter is defined to judge the classifier's performance. Then, a new combination algorithm of multi-classifier is presented. Some properties of combination method are also given in this paper. The Concordia University CENPARMI handwritten digit database is used in the experiment. Nine classifiers which use different features are combined to recognize the image. The experiment results (Recognized,Rejected and Reliability) are 97.05%, 2.05%, 99.08% respectively.
出处
《计算机学报》
EI
CSCD
北大核心
1999年第4期369-374,共6页
Chinese Journal of Computers
基金
国家自然科学基金
国际合作研究项目
关键词
手写体数字识别
多分类器组合
模式识别
Handwritten character recognition, handwritten digit recognition, multi classifier combination
