基于连通域特征的维吾尔手写文本行分割

Connected component feature analysis based handwritten Uyghur text line detection and separation algorithm

针对维吾尔文手写体文本中行分割问题,基于连通域大小将图像中文字分为三类,提出了自适应涂抹细化算法,对主体文本行进行定位;并对第三类连通域中相邻两文本行间粘连的字符进行切割;此外,利用重心范围内的邻域搜索算法,解决了剩余笔画的文本行归附问题。实验结果表明,该方法与常见的水平投影法,分段投影法,及涂抹方法相比具有更好的分割效果。

To deal with the issues of text lines segmentation in Uyghur handwritten documents, based on the size of con-nected components, this paper divides the text image into three categories. In order to get the location of main text-line, it proposes adaptive painting and thinning algorithm. Furthermore, it separates connected characters and assigns them to text lines. In addition, by using the neighborhood search algorithm based on center of gravity, it solves the belonging problems for remaining small strokes. Experimental results show that separately compared with those horizontal projection based, piecewise projection based, smearing based segmentation methods, this method has better text line segmentation results indeed.