摘要： 提出了一种新的用于多字体字符识别的神经网络模型一协同模型，同时提出了一种新的迭代的协同学习算法，该算法将误识率最高的模式的误识样本作为反馈量来修正原型模式．对大量实际采集所得的多字体字符样本的测试表明，该模型能对训练样本集取得１００％的识别率，对多字体字符测试样本的识别也能取得很高的识别率（９８．２％），而且识别速度很快，适用于实时的ＯＣＲ应用．另外，对加噪字符的识别试验表明该模型具有很好的鲁棒性．

关键词: 协同神经网络, 多字体字符识别, 光学字符识别

Abstract: A neural network modal named synergetic model for multifont character recognition is proposed in this paper. At the same time, a new iterative learning algorithm. which modiifies prototype patterns using samples whose recognition rate is the lowest in one loop. The test on the many samples from real applicatioo shows: new model can achieve. 100% recognition rate during training, and tan achieve 98.2% recognition rate in real application. Moreover, its recognition is very quick and fits real-time OCR applicatious. Additionally, the test on noise samples shows the new modal bas strong robustness.