一种基于模板匹配的人民币纸币面额识别方法  

A method for denomination identification of current RMB banknotes based on template matching

作　　者：张国华[1] 梁中华[2] 

机构地区：[1]沈阳工业大学信息科学与工程学院,沈阳110023 [2]沈阳工业大学电气工程学院,沈阳110023

出　　处：《沈阳工业大学学报》2005年第4期439-442,共4页Journal of Shenyang University of Technology

摘　　要：从图像处理的角度论述了人民币纸币的识别方法.针对人民币纸币面额种类少的特点,提出了采用模板匹配的方法.传统的二维模板匹配虽然实现简单,但计算量十分庞大,花费时间太长,针对此情况采用了当今发展比较成熟的一维灰度投影的模板匹配算法.该算法通过图像灰度投影变换,形成一维图像,并用一维图像进行快速匹配.试验结果表明,后者在保证了匹配准确率略高的前提下,识别速度明显高于前者,从而提高了设备的识别速度,使之更具有了实用性.This paper discusses the method of denomination identifying for current RMB banknotes by means of image processing. The approach of template matching is applied according to only few kinds of current RMB banknotes being current. Although the traditional method of two-dimensional template matching can be achieved simply, but the calculating quantities are much larger, it may take very long time. Under this circumstance, algorithm of template matching based on one-dimensional gray projection is adopted. Using this algorithm we get one-dimensional image through gray projection, then use this one- dimensional image to achieve quick match. The result of experiments indicates that the latter method can keep higher matching precision, meanwhile the recognition speed is obviously higher than the former. Consequently, the equipment＇s speed of recognition is greatly increased, which make it have more practicalities.

关 键 词：人民币 图像处理 面额识别 模板匹配 灰度投影 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]