基于凸壳和模糊识别的数学公式识别  

RECOGNITION OF MATHEMATICAL EXPRESSIONS BASED ON CONVEX HULL AND FUZZY RECOGNITION

作　　者：肖建于[1] 王潜平[1] 洪留荣[2] 

机构地区：[1]中国矿业大学计算机科学与技术学院,江苏徐州221008 [2]淮北煤炭师范学院计算机科学与技术系,安徽淮北235000

出　　处：《计算机应用与软件》2008年第5期208-209,254,共3页Computer Applications and Software

摘　　要：根据数学公式中字符或符号间空间关系特点,并针对目前用于数学公式字符空间关系判别的区域和质心方法所存在的不足,提出了基于字符凸壳和模糊识别的字符空间关系判别方法。首先,对数学公式中的字符或符号进行分类,对每一类运用字符凸壳判别其正上和正下关系,然后应用模糊识别的方法对常见关系进行识别。实验结果表明,运用该方法能明显提高字符空间关系判别的识别率,识别的正确率可达到93.5%。This paper proposes a differentiating approach to the spatial relationship of characters based on symbol convex hull and fuzzy recognition theory in consideration of the particularities of such relationship among characters or symbols in mathematical expressions and in accordance with the insufficiency of the region and centroid method currently used for distinguishing the spatial relationships amongst characters in handwritten mathematical expressions. The characters or symbols in the mathematical expressions were firstly classified, and the above and below relationship was discriminated for each class with symbol convex hull,then the familiar relationship was recognized with fuzzy method. The experimental results on the paper demonstrate that the recognition rate of the spatial relationship of characters was greatly improved by the method with the precision rate of 93.5%.

关 键 词：数学公式识别 关系识别 凸壳 模糊识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] P315.7[自动化与计算机技术—计算机科学与技术]