基于综合纠错的印刷体数学公式识别后处理  

Post-processing for printed formula based on synthetic error correction

作　　者：田学东[1] 王文姣[1] 

机构地区：[1]河北大学数学与计算机学院,河北保定071002

出　　处：《计算机工程与设计》2007年第20期5039-5041,5044,共4页Computer Engineering and Design

基　　金：国家自然科学基金项目(60772073)

摘　　要：目前印刷体数学公式识别系统的输出还存在着误识结果,进行必要的后处理是提高识别率的重要手段。介绍了一种将印刷体数学公式识别结果与公式的语义知识相结合的方法,对其误识结果进行系统的分析,给出了若干条共有的规则及基准转移等方法,进行综合纠错的后处理,从而进一步完善印刷体数学公式识别系统。实验结果表明,该方法能够有效地提高系统识别结果的正确率。At present, the output of the printed mathematical expressions recognition system still exist inaccurate result and suitable post-processing is the important way in order to improve recognition rate. Firstly one kind of method which unified the printed mathe- matical formula structure analysis and semantics knowledge is introduced and then the wrong results are analyzed systematically. The methods of several rules in common and baseline shift are given so that the post-processing based on synthetic error correction is carried on. Thus printed mathematical formula recognition system gets consummate. The experiment indicates that the accuracy of the system is enhanced effectively.

关 键 词：公式识别 结构分析 后处理 规则 基准线转移 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]