自动柜员机纸币号码识别装置的研究  

Development of Paper Currency Number Recognition Device in ATM

作　　者：王焱[1] 卢兆第[1] 刘洋[1] 赵延广[1] 

机构地区：[1]辽宁工程技术大学电气与控制工程学院,辽宁葫芦岛125105

出　　处：《计算机测量与控制》2013年第8期2198-2202,共5页Computer Measurement &Control

摘　　要：为了实现自动柜员机(简称ATM)纸币号码自动识别功能,将FPGA和四个接触式图像传感器(CIS)配合使用完成图像的采集,再由AD9822将采集到的图像转换为数字图像,最后DSP对数字图像进行处理,并采用以三种相互补偿结构特征为核心的算法实现号码识别;通过在实验室模拟ATM的纸钞传送过程,安装识别装置进行图像采集和识别,每秒处理7张,对采集的134张人民币进行识别,数字单独识别程序平均识别正确率为99.4%,字母单独识别程序为96.8%,数字、字母混合识别程序为97.7%,结果表明在模拟ATM环境中,纸币号码自动识别装置能够完成人民币号码的图像采集和识别,并且识别正确率也很高。In order to achieve the paper currency number automatic recognition in Automatic Teller Machine （ATM）, the device was de- signed for collecting the image with a FPGA and four Contact Image Sensors （CIS）, acquiring digital image signals from the AD9822, pro- cessing and recognising the digital image by DSP , the paper currency number recognition algorithm of this device was based on three mutual compensation structure characteristics. By conducting an experiment on simulating paper currency transmission of ATM in the lab, recogni- tion device was installed for image acquisition and recognition, processing 7 piece per second, according to the acquisition of 134 piece o{ RMB identification, the average accuracy of the digital single recognition program is 99.4%, the letter single recognition program is 96. 8%, the mixture identification program of letter and digital is 97. 7%, the experimental results showed that paper currency number automatic recogni- tion device can complete image acquisition and recognition of the RMB number in a simulation environment of ATM, and has a high recogni- tion efficiency.

关 键 词：自动柜员机 人民币 图像采集 号码识别 

分 类 号：TP274.2[自动化与计算机技术—检测技术与自动化装置]