一种基于Delaunay三角化的笔划分割算法

字符笔划在字符识别等领域中存在广泛的应用。针对字符图形的骨架化及笔划提取,本文提出了一种基于动态约束Delaunay三角剖分的算法。该方法首先通过对字符的轮廓多边形进行三角剖分,生成一系列具有拓扑关系的三角形,再根据三角形的三种类型生成各三角形骨架,并连接为整个字符的骨架。然后定义了骨架的平滑度函数及分支长度阈值,消除了畸变分支,并根据得到的骨架进行笔划分割。实验结果表明,该算法具有快速稳定的特点且鲁棒性较好。

An Algorithm for pen-stroke extraction Based on Delaunay Triangulation

Character Skeletons plays an important part in character recognition and so on. To get skeleton and pen- stroke of character contours, this paper presents an algorithm based on Dynamic Constrained Delaunay Triangulation. At first, this method decompose the character' s boundary polygon into Dynamic Constrained Delaunay Triangulation to get its topological feature. Secondly, by generating skeleton line segements of three types of triangles and link them, the skeleton of the whole character is gotten. Finally, after dis- branching the skeleton by calculating smoothness function and branch length threshold, the paper shows how to partition the skeleton into pen- stroke skeleton. Experimental results demonstrate that this method have a low calculation complexity and is robust enough.