工程图扫描图像中基于连通群的分类与识别算法

THE CLASSIFICATION AND RECOGNIZING ALGORITHM BASED ON THE LINKED GROUP IN THE SCANNED IMAGE OF THE ENGINEERING DRAWINGS

本文提出基于连通群的图像分割和分类的基础理论，给出了连通群的分类依据，并将其用于工程图纸的识别，对宇串的方向定位算法作了描述。文中关于连通群的理论分析为解决工程图中字符、数字和图形的正确分割、分类及排除噪声的干扰提供了依据，连通群的迭代算法可准确定位字符串的方向。本研究针对建筑结构图的识别作了试验性研究，矢量化与识别精度较高。

This paper introduces the classification,the image sepnding and its theorem background which based on the linked group. The classical basis of the linked group are given which is used to recognize the engineering drawing papers, and the direction determination algorithm for the character string is also described. The accurae expressing of the lined group is supplied for the solution of the correct classification of the stringdigita as well as the graphics in the engineering drawing, and the iteration algorithIn for the linked group can be used to determine the direCtion of the string.We have made the experiment research for the architecture structure drawing papers, the result has proved the algorithm is effectively