摘要
快速准确的纸币清分在银行业中具有非常重要的意义。清分系统包括纸币图像采集、图像预处理、特征提取及分类器设计等几个步骤,其中分类器设计是核心技术基础。论文提出了一种用于高速纸币清分的人民币识别方法,该方法基于整张纸币的特征提取,采用了基于结构风险最小化的高斯混合模型(GMM)设计识别分类器。实验结果表明,提出的方法取得了较高的识别率。
It is important to classify the paper currency at banks quickly and correctly.The system of paper currency recognition includes image eolleetion,preproeessing,feature extraction and classifier design.A critical step is the classifier design.A real time paper currency recognition method is proposed in this paper,which extracts the features based on the whole paper eurreney.A modified GMM is constructed for the Chinese paper currency recognition.The experiments show that GMM which employs structural risk minimization is a more flexible alternative and lead to improved results for Chinese paper currency recognition.
出处
《计算机工程与应用》
CSCD
北大核心
2006年第13期209-212,共4页
Computer Engineering and Applications
关键词
图像识别
特征提取
高斯混合模型
结构风险最小化
image recognition,feature extraction,Gaussian mixture model,structural risk minimization
