-
摘要: 印刷体字符的字形风格和手写字符的书写风格是非常重要的特性. 本文研究了利用字符字形风格之间的约束关系提高识别率的理论和方法, 提出了以字形风格同现概率为基础的 3 种识别模型, 结合实验结果分析了这些模型的优缺点和适用条件, 结果验证了本文提出的风格约束模型能够有效地提高识别率.Abstract: Typeface style and writing style are important features of printed and handwritten characters, but they are not thoroughly studied for character recognition. In this paper, we study how to improve character recognition accuracy by using style consistence between characters. We propose three methods which are based on the co-occurrence probability of styles, and we analyze these methods' features and their suitable conditions with experiments. The result shows that our style consistent models can improve recognition accuracy effectively.
-
Key words:
- Character recognition /
- style /
- co-occurrence probability /
- template matching
-
计量
- 文章访问数: 2416
- HTML全文浏览量: 97
- PDF下载量: 1827
- 被引次数: 0