基于改进Hough变换的符号线段特征提取

Extraction of symbol line-features based on improved Hough transformation

研究了数学表达式识别中符号线段特征的提取问题。为正确提取符号的线段特征及其一些相关属性,提高识别系统的符号识别率,对Hough变换的局限性进行了分析,并作了一些改进,改进算法在原算法的基础上增加了线段的连续性的检测以及相关属性的提取,如线段数目、方向(角度)以及始末端点等。实验证明,该算法能有效地提取出符号线段及其一些属性,可应用于其他识别系统的符号特征提取。

In order to correctly extract the line-features of symbols and some of their attributes, Hough transformation has been improved based on the analysis of its limitations, and it has therefore the additional inspection of line continuity and extraction of associated attributes, such as number, direction (angle), start and finish of lines. Experimental results show that the algorithm proposed can be used to extract symbol lines and some of their attributes, and it can be used for extraction of symbol features in other recognition systems as well.