基于决策导向非循环图支持向量机的脱机手写体汉字识别  

Recognition of Handwritten Chinese Characters Based on DDAG SVM

作　　者：吴仕莲 杨杰[1] 赵冬琴[1] 

机构地区：[1]南京航空航天大学自动化学院,江苏南京210000

出　　处：《数字技术与应用》2016年第7期41-42,45,共3页Digital Technology & Application

摘　　要：目前脱机手写体汉字识别在小字符集方面取到了比较好的效果,但在大字符集方面仍存在着识别速度慢、准确率低等问题。不同于传统的二叉树方法,本文将决策导向非循环图用于汉字识别,并加以改进。仿真实验表明,该算法能对大字符集的手写体汉字进行识别,有效减小了误差,具有较高的识别率。Currendy,off-line handwritten Chinese characters recognition in the small category character recognition has obtained good effect. However, recognition in large character set still has the weakness of low efficiency and low accuracy.This paper presented an improved SVM-based algorithm of Decision Direct Acyclic Graph（DDAG）. Simulation results demonstrate that the algorithm can recognize large character set of handwritten Chinese character and has high recognition performance.

关 键 词：支持向量机 多分类 脱机手写体汉字 决策导向非循环图 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]