本文论述并设计实现了一个脱机自由手写体数字识别系统。文中首先对待识别数字的预处理进行了介绍，包括二值化、平滑滤波、规范化、细化等图像处理方法；其次，探讨了如何提取数字字符的结构特征和笔划特征，并详细地描述了知识库的构造方法；最后采用了以知识库为基础的模板匹配识别方法,并以MATLAB作为编程工具实现了具有友好的图形用户界面的自由手写体数字识别系统。实验结果表明，本方法具有较高的识别率，并具有较好的抗噪性能。-This paper describes and designs a free handwritten number recognition system. Firstly,

the pretreatment of the character to be recognized is introduced, including binarization,　smoothing, normalization and thinning. Next, how to extract the structural features of the numbers is discussed, and we describe the constructing method of repository. Finally, we use the method of template matching, based on repository, to recognize the digital number. Matlab is used as a program tool to realize this free handwritten digital recognition system with friendly graphical user interface. The experimental results show that the rate of the recognition system is high, and the proposed method is robust to nois