摘要
根据多分类器组合原理,提出一种基于神经网络的多分类器组合模型.该模型首先使用基于贯穿码特征的分类器对字符分类,然后,由基于投影特征的分类器对经过上一级分类器分类后的字符进行识别.试验表明,该模型能有效提高光学字符识别率.
According to the theory of multi-classifier combination, a combination model for OCR（Optical Character Recognition） based on neural network is given. A primary classifier is utilized to classify the character by cross feature code before the secondary classifiers with projection feature are applied to further recognizing the characters. The experiment results show that the model can increase the OCR rate effectively.
出处
《武汉理工大学学报（交通科学与工程版）》
2007年第6期1110-1112,1116,共4页
Journal of Wuhan University of Technology（Transportation Science & Engineering）
基金
海军工程大学科研基金项目资助(批准号:HGDJJ04E364)
关键词
光学字符识别
多分类器组合
BP神经网络
特征提取
optical character recognition
multi-classifier combination
BP neural network
feature extraction
