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。-In this paper, designed and implemented an off-line handwritten numeral recognition system. The paper first pre-treatment identification numbers were introduced, including binarization, smoothing filter, normalization, thinning and other image processing methods　secondly, to explore how to extract the number of characters in the structural characteristics and stroke features, and described in detail Knowledge of construction methods　finally adopted in order to Knowledge-based template matching recognition method, and to MATLAB as a programming tool to achieve a friendly graphical user interface handwritten numeral recognition system. The experimental results show that this method has higher recognition rate, and has good anti-noise performance.