基于机器学习的手写汉字特征选择  

Feature Selection of Handwritten Characters Based on Machine Learning

作　　者：钱国良[1] 舒文豪[1] 洪勇 叶风[1] 郭茂祖[1] 

机构地区：[1]哈尔滨工业大学计算机科学与工程系计算机应用软件教研室

出　　处：《哈尔滨工业大学学报》1998年第1期57-60,共4页Journal of Harbin Institute of Technology

摘　　要：根据机器学习理论，提出了在手写汉字识别中，特征提取与识别同时进行的方法，并将决策树归纳学习算法ＩＤ３成功地应用于学习与识别过程。在识别过程中，依据决策树有选择地提取特征进行匹配，减少了特征提取的数量，大大地提高了识别速度。Based on the theory of machine learning a new approach of feature selection and recognition working simultaneously on handwritten Chinese characters is proposed, and the algorithm for decision tree induction -ID3 is successfully used in feature selection and recongnition. In the process of recognition, it extracts some useful features for matching, reduces the number of extracted features and improves recognition speed greatly.

关 键 词：特征选择 机器学习 手写体 汉字 模式识别 

分 类 号：O235[理学—运筹学与控制论]