Abstract: According to the characteristics of handwritten letter recognition, a new handwritten letter recognition algorithm is proposed based on the rough set theory. The hypothesis of normal distribution has been tested to ensure the reliability of the collected samples. By using upper approximation, lower approximation, as well as positive region of rough set, feature selection is made in order to simplify the decision-making system. Classification rules are then extracted from the simplified decision-making system. Experimental results show that the new algorithm has higher recognition accuracy, and is feasible and effective.

Key words: rough set, positive region, handwriting recognition, feature selection, supervised learning

摘要： 针对手写字母识别的特点，结合粗糙集相关理论，提出了一种新的手写字母识别算法。通过对采集的样本进行正态分布假设验证，保证样本的可靠性；利用粗糙集上近似、下近似以及正域概念，对手写样本决策系统进行特征选择以简化决策系统，并进一步提炼手写分类规则。实验结果表明，新算法具有较高的识别准确率，是有效可行的。

关键词: 粗糙集, 正域, 手写识别, 特性选择, 监督学习