一种识别手写汉字的多分类器集成方法

根据多信源信息处理与字符识别的经验知识，提出了一个识别手写汉字的多分类器线性集成模型．这个模型不仅考虑到不同的分类器对不同字符识别能力的不同，而且还考虑了不同的分类器得出的输入字符与参考模板之间相似度的实际大小对判决的影响，及不同分类器提供的候选字符对判决的支持作用，更重要的是提供了一种通过监督学习，利用计算机程序自动计算模型参数的方法，因而实现了一个较好的集成系统．同时，本文还提供了三个用于集成的分类器，它们集成的结果充分显示了本方法的有效性。

A METASYNTHETIC APPROACH FOR HANDWRITTEN CHINESE CHARACTER RECOGNITION

As a high level integration approach, metasynthesis has drawn much attention. In this paper, a metasynthetic approach for combination of multiple classifiers is proposed. As a first step, a linear integration model is built. In this model, not only the degree of the similarity between the input and its ranked candidates are appled, but also the supporting effects of multiple candidates are taken into account, and their contributions to the integration result is expressed by a linear function. Then, an algorithm for automatically calculating the arguments of the model through supervised learning is provided. Experiments on metasynthesis of three individual classifiers for handwritten Chinese character recognition are given to demonstrate the effectiveness of our method.