基于笔划组合的维吾尔文联机手写单词连体段分割算法

ALGORITHM OF SEGMENTING CONJOINED SECTION OF ONLINE HANDWRITTEN UYGHUR WORDS BASED ON STROKE COMBINATION

从维吾尔文的特征和书写规则出发对维吾尔文联机手写单词识别技术进行了探索性研究,并提出一种新的思路:不是直接把单词切分成字母,而是先把单词分割成连体段,然后再分割成字母。这样,可以提高字母切分和字母识别的准确率。按照该思路,提出一种连体段分割算法:根据通过研究维吾尔文的特征和书写规则找出来的一些规则把一个个的笔画,组合成连体段。实验证明了该连体段分割思路和算法的可行性。

In this paper,based on Uyghur structure characteristics and writing rules,we study the recognition technology of online handwritten Uyghur words,and put forward a new idea,that is the handwritten words will be segmented into conjoined sections first and then further cut the sections into single letters rather than to directly cut the words into letters.In this way the letter segmentation and recognition accuracy can be improved.According to this idea,we present a conjoined section segmentation algorithm.It combines individual strokes into conjoined section based on some rules which are discovered from the research on Uyghur structure characteristics and writing rules.Experimental results approved the feasibility and accuracy of the conjoined section segmentation algorithm presented in this paper.