For the problems of the traditional text lines extractions, this paper proposed a new method to extract multi-oriented text lines from offline handwritten Nushu characters image. Firstly, by morphological dilating and minimum enclosing rectangle, it extracted blocks of Nushu characters from the preprocessed Nushu characters image. It built a triangulation network by Delaunay triangulation on these blocks. According to three given rules,it constructed a link model by calculating the weight of each edge of the triangulation network. Finally,it extracted the optimal text lines by the ordered edges and mutual exclusion. The experimental results show that this method can effectively extract the multi-oriented text lines from Nushu characters images on different carriers, such as fans and handkerchiefs.%针对传统文本行提取方法的不足,提出一种脱机手写女书字符图像多方向文本行提取方法.该方法先将预处理后的女书字符图像通过形态学膨胀和最小外接矩形来提取出女书字符块,然后用Delaunay三角剖分对字符块构建三角网,再根据三个给定的准则计算三角网中每条边的权重构建链接模型,最后根据依权重排序后的边,利用互斥原则搜索出最优文本行.实验结果表明,该方法能有效提取扇面、布帕等不同载体上女书字符图像中的多方向文本行.
展开▼
