三枝粗糙集和变粒度原理的手写体汉字识别  

Study of handwritten Chinese characters recognition based on three branches rough sets and variable granularity principle

作　　者：王建平[1] 王梦泽[1] 

机构地区：[1]合肥工业大学电气工程及自动化学院,合肥230009

出　　处：《计算机工程与应用》2014年第22期223-227,共5页Computer Engineering and Applications

摘　　要：将三枝决策粗糙集模型用于手写体汉字识别,建立三枝决策粗糙集模型,其区域分类以正、负和边界为基础,更好地体现手写体汉字识别分类近似性。定义了脱机手写体汉字识别决策信息系统特征属性约简相对粒度熵和属性重要度,将手写体汉字识别从属性重要度的角度出发,设计一种基于粒与粒之间逻辑运算的自下而上的粒度网结构,使手写体汉字识别属性集的适用粒度逐渐变大,可以获取更多的有效信息。仿真实验表明,该方法是可行有效的。The model of the theory of three-way decision-theoretic rough sets is used for the recognition of handwritten Chinese characters. A model of the theory of three-way decision-theoretic rough sets is set up which the regional classifi-cation is based on positive, negative and boundary. It can better reflect the recognition of handwritten Chinese characters classification on approximation. And define characteristic attribute of reduction of relative granularity entropy and the sig-nificance of attribute in offline handwritten Chinese characters recognition decision information system. If consider the recognition of handwritten Chinese characters from the point of view of the importance of attribute, it can design a kind of bottom up particle network structure based on logical operation between grain and grain, you can obtain more effective information. Simulation results show that, the method is feasible and effective.

关 键 词：粗糙集 三枝决策 风险损失 决策约简 粒度分层 

分 类 号：TP391.12[自动化与计算机技术—计算机应用技术]