多字体字符识别的分级协同模型

提出了一种用于多字体字符识别的分级协同神经网络模型。该分级模型的识别由两部分组成：Haken的协同网络确定识别进入哪个协同子网；用协同子网进行具体识别。对大量实际采集得到的多字体字符样本的测试表明：新模型能有效地提高协同神经网络对多字体字符的识别率，但由于仍保留了识别速度快的特点，所以新模型适用于实时的光学字符识别应用，对加噪字符的识别试验表明该模型具有很好的鲁棒性。

Hierachical Synergetic Model of Multifont Character Recognition

In order to improve the recognition performance of synergetic neural network further, a hierachical synergetic neural network model for multifont character recognition was proposed. The recognition of hierachical model is composed of two steps. First, Haken's synergetic neural network decides which synergetic subnetwork the test sample should be input in, then the detailed recognition is carried out by synergetic subnetwork. The test on the many samples from real application show that the new model can improve recognition rate of multifont character recognition effectively. Moreover, its recognition is very quick and fits real time optical character recognition(OCR) applications. Additionally, the test on noise samples shows the new model has strong robustness.