摘要
通过调研国内外纸币面额识别技术的研究现状,设计了一种基于紫外荧光图像的人民币面额识别算法,该算法利用在紫外光照射下人民币正面的荧光面额区域作为识别特征,经过简单的图像预处理(包括倾斜校正、图像二值化、滤波和形态学处理等)操作,综合运用几何结构分析和模板匹配的方法实现对六种人民币面额的识别,识别率达到99%以上,平均识别时间在0.04s以内。该算法识别效果好并且满足一定的实时性,经过改善可以运用到自动售货机、点钞机、清分机、ATM机等实际系统中。
By investigating the current research state of denomination recognition for banknotes at home and abroad, designed a denomination recognition algorithm for RMB based on image processing of UV fluorescence zone. This algorithm uses the fluorescence zone of RMB under the UV light as recognition feature, after simple image pre-processing operations （including skew correction, image binarization,filtering and morphological processing, etc.）, and combines the geometry analysis and template matching to recognize six types of RMB. The correct rate of this algorithm can reach 99% and the average recognition time is less than 0.04s .after testing. In general, this algorithm is effective and satisfies the real-time requirement. After some improvement, it can be applied to the actual system.
出处
《机械设计与制造》
北大核心
2017年第3期1-3,共3页
Machinery Design & Manufacture
关键词
纸币识别
面额识别
纸币清分
图像识别
Banknote Recognition
Denomination Recognition
Banknote Sorting
Image Recognition
