摘要
提出了一种基于特征区域投影特征匹配的钞币面值识别算法。针对不同面额的人民币,所选取的特征区域位置、大小和数目可不同。建立投影特征模板库,并采用分类比较和滑动匹配的方法提高算法速度和适应能力。实验结果表明:利用该算法对人民币的识别速度达到2000张/min,可满足嵌入式系统对算法实时性的要求。
A banknote denomination recognition algorithm based on characteristic area projection matching is proposed. The selected characteristic area position, size and number against banknotes with different denominations can be different. A template database for projection characteristics is established. Operating speed and adapting capacity are improved by using sort comparison and slide matching method. The experimental results show that the recognition speed for banknotes with this algorithm can be up to 2000 pieces/min and this algorithm can fulfill the real time demands of the embedded system.
出处
《光电工程》
EI
CAS
CSCD
北大核心
2004年第1期59-61,共3页
Opto-Electronic Engineering
