手写体汉字识别问题属于一种大规模的模式识别问题。本文基于球邻域模型的几何意义解释,即将神经网络的训练转化为几何的点集覆盖问题,通过对神经网络分界面的分析提出了一种改进的前馈神经网络训练算法,并且引入神经网络集成的思想,用以解决手写体汉字的识别问题,实验结果表明该算法可以用来解决大规模的模式识别问题且具有较好的效果。 Handwritten Chinese character recognition belongs to a large-scale pattern recognition problem. Based on the geometric meaning of the ball neighborhood model, this paper transforms the neural network training into the point set covering by geometry. An improved feedforward neural network training algorithm is proposed based on the analysis of the neural network interface. The neural network training Integrated ideas to solve the problem of recognition of handwritten Chinese characters. The experimental results show that the algorithm can be used to solve large-scale pattern recognition problems and has good results.