基于综合特征的字符模板库的建立与训练

Building and Training of Character Template Litrary Based on Synthetical Features

叙述邮政编码自动分拣系统的软件处理的原理以及程序设计方法,对于倾斜放置信封的邮编区域的定位问题提出解决方案,利用信息熵理论对字符的综合特征的选取提出融合方案,建立训练了邮政编码识别系统中的手写数字模板库,并设计与之相关的可视数据库系统,该系统大大提高信封的自动分拣率以及可行性。

This paper describes the principles and software design technologies of automatic sorting postal code system. It proposes a solution to locating the region of a declined envelop and makes an integration program for selecting synthetically feature of character by using the information entropy theory. It establishes and trains a handwriting digital fea- ture library of automatic sorting postal code system. It also designs a reasonable and related database system which greatly increases the recognition rate of the envelope and the feasibility of automatic sorting postal code system.