数学公式图像的结构理解与重现  

Reconstructing mathematical expressions from image data

作　　者：史广顺[1] 肖萃[1] 王庆人[1] 

机构地区：[1]南开大学机器智能研究所,天津300071

出　　处：《智能系统学报》2008年第5期401-407,共7页CAAI Transactions on Intelligent Systems

基　　金：天津市自然科学基金资助项目(05YFJMJC01500)

摘　　要：数学公式图像识别与理解是文档图像处理领域的重要组成部分,目前尚无满足一般应用的处理方法.提出了一种鲁棒的数学公式结构理解方法,使用公式图像识别结果、语法规则和句法规则分析数学公式结构,对数学公式的类型进行了完整的划分,对识别结果的错误进行自动的检查和纠正,能够自动分析数学公式符号的优先级和计算顺序.既可以应用于数学公式图像的识别与格式转换,也可应用于对数学公式的检索和辅助编辑.基于1 000个真实公式图像的实验结果证明了分析方法的有效性和稳定性.Mathematical expressions appear in many kinds of scientific documents and technical reports. Understanding and reconstructing mathematical expressions has become an important problem in the domain of document image analysis. The authors developed a robust method for the analysis of structure in mathematical expressions. After images are processed, generating recognition results, this method analyzes the structure of mathematical expressions according to syntax rules and syntactic rules. Classification into different types of mathematical expressions is then made. Syntax errors in the recognition process are checked and corrected automatically. The preferential level and the computing sequences of arithmetical operation signs in mathematical expressions are also automatically analyzed. This method can be applied to the recognition of images containing mathematical expressions and transforming between formats, and is useful in retrieval and editing of mathematical expressions. About 1000 images of mathematical expressions from real documents were used for performance evaluation. The test results proved the stability and efficiency of this method.

关 键 词：数学公式识别 版面结构分析 语法结构分析 数学公式结构理解 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]