一种基于粗糙集理论的票据图像识别分类方法

A Coupon Images Classification Method for Recognition Based on Rough Set Theory

本文根据票据图像的版面特点,选取表格线、背景信息及纹理信息作为主要特征,采用粗糙集理论进行特征属性约简、规则提取及规则约简。并提出了一种基于多规则集的综合决策分类方法,巧妙的利用其分类及拒识的特点,有效地实现了票据图像分类。

In this paper, we first select coupon image grid, background information and texture information as main features, then we use Rough Set Theory to reduce feature properties, at last we get useful feature vectors and rules for cluster coupon images. We also produce a comprehensive decision classification schema based on multi-rules sets. This schema makes a good thing out of success and failure of recognition and accomplishes the classification of coupon images effectively.