改进的决策树算法在手写体汉字识别中的应用  

Application of improved decision tree algorithm in recognition of handwriting Chinese character

作　　者：施梅芳[1] 贲黎明[1] 

机构地区：[1]常熟理工学院计算机科学与工程学院,江苏常熟215500

出　　处：《兰州理工大学学报》2011年第4期106-109,共4页Journal of Lanzhou University of Technology

摘　　要：介绍脱机手写体汉字识别的基本流程,以传统决策树ID3模型为基础,对决策属性挑选策略进行改进.该方法基于信息增益的计算标准,通过引入属性间相关性关系对传统ID3算法"多值偏向性"的缺陷进行补偿,提出节点协调度的预剪枝策略.从分类精确度方面对改进的ID3算法和传统ID3算法进行比较和分析.将改进的ID3算法应用于手写体汉字识别中,试验结果表明,改进的ID3算法可以减少训练时间,并提高识别率.Basic flow of off-line handwriting Chinese character recognition was introduced.Then,taking the traditional decision tree ID3 model as basis,the selection strategy of decision attributions was modified.Based on the computation standard of information gain and by introducing correlativity of the attributions,the defect of ＂multiple-valued bias＂ of traditional ID3 algorithm was supplemented by using this method.A pre-pruning strategy was proposed based on node coordination degree.The classification precision of the improved ID3 algorithm and the traditional ID3 algorithm was analyzed and compared with each other.Finally,the improved ID3 algorithm was used for handwriting Chinese character recognition.Experimental result showed that the improved ID3 algorithm could reduce training time and improve recognition percentage.

关 键 词：决策树算法 决策属性选择 预剪枝策略 手写体汉字识别 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]