【摘要】 为解决联机手写汉字识别技术中兼顾连笔和笔顺的难题，提出了联机手写汉字的模糊属性关系图描述和基于知识的松驰匹配方法。该方法能充分描述汉字结构、对手写变形不敏感，并具有良好的连笔字识别能力和适应不同笔顺的特点。基于上述方法，作者实现了一个联机手写汉字识别系统。实验结果表明，本方法能够有效地增加对于低限制书写汉字的识别正确率，并具有较好的抗噪声能力。经过２２５３０汉字测试，首位识别率达９８．８％，前十位识别率达９９．７％。更多还原

【Abstract】 ? In order to solve the problem in on line Chinese character recognition that how to deal with strokes connection and stroke order variation, this paper presents a novel description of on line handwritten Chinese character using a fuzzy attributed relational graph(FARG), as well as a knowledge based relaxation algorithm for FARG matching. This method has the charateristics that, it can fully utilize the structure feature of Chinese characters, and not sensitive to writing variation. It is not only able to recognize connected script, but also can it tolerate different stroke writing order. An on line Chinese character recognition system is built up based on the aforementioned principle. The experiments show that, our approach is effective in improving the recognition rate of low constraint writing, and robust against noises as well. 22 530 handwritten Chinese characters are used to test our algorithm, a recognition rate of 98.8% is achieved for first candidate, and 99.7% for first ten candidates.更多还原