Abstract: Mathematical expressions recognition has two main processes. One process is symbol recognition and another process is structural analysis. Most existing approaches do structural analysis after symbol recognition. They can easily fail when symbol identify errors occur. Moreover, they assume the availability of the whole mathematical expression before being able to analyze the semantic information of the expression. To tackle these problems, this paper proposes a real time recognition approach for dynamic recognition of handwritten mathematical expressions. The proposed approach based on a combination of symbol recognition and structural analysis constructs a mathematical expression tree to recognize mathematical expression. In the process of constructing mathematical expression tree, it uses a method of positioning the affected area eliminating duplication of other unaffected regions identification, thus increasing the efficiency of re-identification and having the advantage of allowing disorder input. Experiments show that the approach is able to achieve very good performance results.

Key words: tree of mathematical expression, real-time recognition, positioning the affected area, structural analysis

摘要： 数学表达式识别一般分为字符识别和结构分析两部分，而且大多数现有的方法是先进行字符识别然后将字符识别的结果作为结构分析的输入再进行结构分析，在这种分步识别的过程中，字符识别的错误会被继承到结构分析阶段，最终导致识别错误。关于数学表达式结构分析的问题，现有的方法大多是在假设所有的符号已经识别的基础上进行的。为了解决上述问题，提出了一种实时识别联机手写数学表达式的方法。该方法基于字符识别和结构分析的结合，动态地构建一棵数学表达式结构树来识别该数学表达式。在构建数学表达式树的过程中，采用了影响区域定位的方法，免去了其他不受影响区域的重复识别过程，因而提高了再次识别的效率，同时还弥补了现有实时识别方法不能乱序输入的缺陷。实验结果表明提出的方法可以得到比较满意的识别结果。

关键词: 数学表达式树, 实时识别, 影响区域定位, 结构分析