摘要
对于经典Rough集理论中某一类决策或模式识别问题,其样本空间或决策表中客观存在的任何两个样本实例体现的决策规则不可能完全相同。本文对此进行了讨论并提出了RS邻域拓展的全局补偿RS方法。基于上述方法对脱机手写识别英文字母的模板匹配算法进行了优化,说明了其有效性。
This paper discusses a kind of problems of decision or pattern recognition in classical rough set theory, which cannot have identical decision rules for any two samples in sample spaces or decision tables. A method of neighborhood expanding Rough Sets is proposed. Based on the method, the template match algorithm is optimized and effectively applied to off- line handwritten English character recognition.
出处
《南昌大学学报（理科版）》
CAS
北大核心
2005年第6期624-627,共4页
Journal of Nanchang University(Natural Science)
基金
江西省自然科学基金资助项目(0311101)
南昌大学科学基金资助项目(Z-02951)
关键词
全局补偿
邻域拓展
决策规则
模板匹配
脱机手写识别
AGENT
global compensation
neighborhood expanding
decision rules
template match
off - line handwritten recognition
Agent
