一种有效的手写汉字多步分割方法

在字符识别系统中,字符的有效分割是识别的关键。针对手写汉字字间距及字内距无规则可循,字符间极易发生粘连、交错等现象,提出一种多步分割方法。该方法首先利用Viterbi算法将原字符串切分成互不连通的分割块,使非粘连汉字、交错汉字得到正确分割;对于其中宽度较大存在粘连字符的分割块,从候选分割点入手,用非线性分割路径将粘连部分分开;最后再应用A*算法找到全局最佳分割位置,使过分割的字符得到完整合并。实验结果表明,该方法对于手写汉字的分割是可行、有效的。

An Effective Multi-stage Segmentation Method for Handwritten Chinese Characters

Handwritten Chinese charaters segmentation is crucial to their recognition in OCR system.In practice,handwritten Chinese characters may be written touching or overlapping each other,and between-character gap or inter-character gap has large variations.For the problem posed by above situations,a multi-stage method of handwritten Chinese characters segmentation is proposed.Firstly,Viterbi algorithm is employed to obtain the no touching or overlapping characers from the separated blocks of the characters string.Then,for the touching charcters in the wide blocks,the segmention is performed by searching non-linear segmentation paths which pass through the candidate segmentation points.Finally,A* algorithm is applied to determine the optimal segmention paths.As a result,over-segmentation components are merged into whole characters.Experimental results show the proposed method is feasible and effective for off-line handwritten Chinese characters segmentation.