【摘要】 人主要是依靠汉字的结构特征来识别汉字,本文提出用汉字笔画端点、折角点和交叉点以及汉字笔画之间构成的"洞"的中心点在24个区域上的分布直方图作为汉字结构新特征,提出通过对Harris角点的检测、合并和去除操作的汉字笔画端点、折角点和交叉点的提取方法,通过建立手写体与印刷体汉字的特征库中相同汉字新特征的集聚性与不同汉字新特征间的离散性的计算,从实验上验证了所提出的汉字新特征的有效性,分析了所提出的新特征对汉字的大小、字体、光照等影响具有良好的鲁棒性,该新特征对手写体汉字的识别具有较好的应用价值。更多还原

【Abstract】 People mainly rely on the structural characteristics of Chinese characters to identify Chinese characters,the center point is proposed in this paper Chinese characters stroke endpoints,composed between angle point and cross point and Chinese characters strokes "hole" in the 24 regional distribution histogram as a new feature extraction method is proposed Chinese characters structure,the extraction method of Harris corner detection,merging and removal operation Chinese characters stroke endpoints,angular point and cross point,through the establishment of handwritten and the printed Chinese characters feature library,computing discrete between the new characteristics of different Chinese characters and agglomeration of new feature the same Chinese characters in the feature library.From the experiments,the proposed new features Chinese characters the effectiveness,analysis of the impact of the new features of the proposed Chinese characters font,size,light has good robustness,this new feature has a good application value for the recognition of handwritten Chinese characters.更多还原