汉字的笔段网格字形描述及字形比对算法
Stroke-Segment-Mesh Depiction of Chinese Character Glyph and Algorithm for Glyph Comparing
-
摘要: 由于目前计算机缺乏对汉字字形统一有效的形式化描述和比对计算方法,致使无法描画输入所需的各种可能汉字,也无法利用计算机对字形进行比对分析.提出一种具有颗粒度适当、无歧义、规范化基元,能描述各种可能字形(包括错字、古籍异体字、拼合字)骨架异同的笔段网格汉字字形描述方法;并基于该方法给出了字形比对算法,它能自动提取字形包含的简单笔画和复合笔画,根据字形的不同自适应地选取复合笔画或简单笔画,并以此为单位进行比对;最后将计算两字形最优配对笔画间的向量距离总和作为比对结果.实验结果表明,该方法具有很强的字形描述能力,字形比对算法对结构规范字形的比对准确率较高,可用于支持各种汉字的描画输入及面向字形比对分析的各种应用.Abstract: Due to the lack of a universally applicable yet effective method of formal description and automatic comparison for Chinese character glyph,we can hardly input arbitrary Chinese characters by glyph drawing,nor can glyph comparison and analysis be realized with algorithms. A stroke-segment-mesh description of Chinese character glyph is presented. The stroke elements of a glyph are standardized,with suitable size,capable of depicting the glyph skeleton of any Chinese characters (including wrongly written characters,variant forms of characters in ancient literatures,and combined-characters). Based on this description,a glyph comparing algorithm is developed,which can automatically decompose a glyph into simple strokes and compound strokes,adaptively select either of them for comparison according to the shape of the glyph. It then calculates the sum of the vector distance of best-matching strokes of the two candidate characters as the comparing results. Experimental results indicate the versatility of this description and the high accuracy of the proposed algorithm. Our new approach provides support not only to Chinese character input but also to all kinds of glyph-analysis-oriented applications.
下载: