一种基于离散Voronoi图的手写体文字细化方法

Algorithm for thinning hand-written text based on Voronoi diagram

基于一般图形Voronoi图的离散构造法,提出了一种新的文字图像细化算法。该方法首先对文字图像进行水平扫描和垂直扫描,通过游程匹配记录下所有端点游程,并对端点游程进行处理,接着计算文字图像的边界,在计算边界的同时根据端点游程划分出生成元。最后基于一般图形Voronoi图的离散构造法生成文字图像内部的Voronoi边,从而得到文字图像骨架。该方法直接从图像的边界入手,解决了当前已有算法从图像边界近似多边形入手的问题。该方法速度较快,尤其在大篇幅文字图像的细化速度方面具有显著优势,且简单易行,可以较精确地获取文字图像的骨架。

Based on Voronoi diagrams,the authors give a new method of thinning text image.The algorithm scans the text image horizontally and vertically,memorizes all the ends of line segment through the result of matching line segment at first,and deals with the ends of line segment.Then calculate text image boundary,plot generator based on end of line segment at the same time.Finally,according to Voronoi diagrams’ discrete method,the authors get Voronoi border,which is the skeletons of hand-writing text.It is based on image boundary straightly,solves algorithms’ problem formerly which is based on approximate polygon of region boundary.The algorithm has the advantages of fast speed,especially in image at great length.And the algorithm is accurate,efficient and easy to perform.