摘要： 提出了一种基于神经和模糊方法的手写体邮件地址的识别方案，并详细分析了神经和模糊方法应用于手写体识别问题的3种技巧：指派字符可信度值，给出字符分段上下文信息，利用模糊积分汇总字符的可信度值。

关键词: 手写体识别, 神经网络, 模糊逻辑, 多层前馈神经网络, 模糊积分

Abstract: This paper presents a mail address recognition technology based on neural and fuzzy methods. We analyse threetechniques for applying neural and fuzzy methods to handwritten address recognition problem: assigning character confidence values, giving the segments contextual information and aggregating character confidence values by means of fuzzy integrals.

中图分类号: 

• TP391.4