摘要
纸币面额识别是纸币清分过程中的一个很重要的步骤,目前对纸币面额识别的研究主要集中于灰度图像上,还没有人使用彩色图像做过纸币面额识别的研究。在此提出一种基于HSV空间的纸币面额识别算法,根据待识别纸币和不同面额纸币的H、S、V分量均值的色差大小来对纸币的面额进行识别。在此首先使用扫描仪扫描到的高分辨率的图像进行实验,证明了算法的正确性。然后采用智能点验钞机采集到的低分辨率的图像进行实验,针对低分辨率图像,提出采用直方图均衡化的改进算法,分类准确率达到98.78%,证明本算法能够满足实际应用的需要。
Denomination recognition of banknote is a very important step in the process of paper currency sorting. The cur-rent research of the denomination recognition of banknote is mainly concentrated in the gray image,but no one uses color image for the research. A banknote denomination recognition algorithm based on HSV space is presented in this paper, which recognizes the denomination of banknote according to the color scale of H,S,V component of different banknotes. The correctness of the algorithm was proved by the high resolution images got by a scanner. The low resolution images collected by the smart currency detector were employed to carry out an experiment for recognition of low resolution images. An improved algorithm of using histo-gram equalization is proposed,whose classification accuracy rate reaches 98.78%. It proves that this algorithm can meet the needs of practical application.
出处
《现代电子技术》
北大核心
2015年第2期88-91,95,共5页
Modern Electronics Technique
关键词
纸币清分
面额识别
HSV空间
色差
paper currency sorting
denomination recognition
HSV space
color difference