摘要
应用于点钞机的纸币冠字号识别系统工作过程可分为图像采集和冠字号识别两大阶段。目前,已有的纸币冠字号识别系统的在线识别速率最快为15张/s,难以和主流的出钞率为20张/s的点钞机同步工作,其主要原因是系统采集纸币图像耗时过长,而预留给冠字号识别的时间太短。针对该问题,并结合该研究现有的冠字号识别算法研制出一款基于DSP+CPLD的高速纸币图像采集系统。将该系统安装在出钞速率为20张/s的点钞机中,实际运行结果表明,此系统可连续完整地采集任意面向的纸币图像,完整采集一张纸币的时间为26 ms,而一张纸币在点钞机中的时间为50 ms,这就为识别阶段预留了充足的时间,保证了识别系统的实时性。
The currency notes number identification system is including currency image acquisition unit and image recognition unit. The highest speed of the current currency notes number identification system is 15 pieces per second, and the identification rate is hard to synchronize with the rate cash register which dispensing speed is l 200 pieces per minutes. The main reason for that is acquisition for one cash costs much time,means that limited time was left for image recognition. For this problem, a high speed image acquisition method based on DSP and CPLD has been proposed and a currency notes number identification system based on cash register has been developed. The currency image acquisition system is installed in a kind of cash register which dispensing speed is 1200 pieces per minutes, its running result shows the whole paper currency image can be continuously collected and the time for collecting one is 26 ms, and the time for one note remain in the cash register is 50 ms, so it can reserved sufficient time to ensure the real-time property of the system.
出处
《仪表技术与传感器》
CSCD
北大核心
2013年第11期22-24,27,共4页
Instrument Technique and Sensor
基金
沈阳市科技计划项目(F10-213-1-00)
关键词
点钞机
纸币
采集
DSP
CPLD
cash register
paper currency
acquisition
DSP
CPLD