高光谱成像的纸币真伪鉴别研究  

Identification of True and Counterfeit Banknotes Based on Hyperspectral Imaging

作　　者：张振清 董丽娟[2] 黄宇 陈兴海 黄威[5] 孙勇[6] ZHANG Zhen-qing;DONG Li-juan;HUANG Yu;CHEN Xing-hai;HUANG Wei;SUN Yong(Department of Criminal Science and Technology,Railway Police College,Zhengzhou 450053,China;Shanxi Provincial Key Laboratory of Microstructure Electromagnetic Functional Materials,Shanxi Datong University,Datong 037009,China;Institute of Public Safety Research,Zhengzhou University,Zhengzhou 450001,China;Wuxi Spectrum Vision Technology Co.,Ltd.,Wuxi 214000,China;Institute of Forensic Science,Ministry of Public Security,Beijing 100038,China;MOE Key Laboratory of Advanced Micro-Structured Materials,School of Physics Science and Engineering,Tongji University,Shanghai 200092,China)

机构地区：[1]铁道警察学院刑事科学技术系,河南郑州450053 [2]山西大同大学微结构电磁功能材料山西省重点实验室,山西大同037009 [3]郑州大学公共安全研究院,河南郑州450001 [4]无锡谱视界科技有限公司,江苏无锡214000 [5]公安部物证鉴定中心,北京100038 [6]同济大学物理科学与工程学院先进微结构材料教育部重点实验室,上海200092

出　　处：《光谱学与光谱分析》2022年第9期2903-2912,共10页Spectroscopy and Spectral Analysis

基　　金：国家重点研发计划项目(2017YFC0803806,2018YFC0807203-3);国家自然科学基金项目(91850206,11974261);山西省科技创新团队项目(201805D131006);山西省重点研发计划项目(201903D121026);河南省科技攻关项目(192102310515,212102310387)资助。

摘　　要：人民币是中华人民共和国的法定货币,人民币的真假直接关系到我国社会的和谐与稳定。百元钞票作为我国最大面额钞票,其真假的鉴定显得更为重要。以2005版的百元钞票为例,利用可见/近红外高光谱成像仪获取一张真钞与两张假钞正、背面高光谱影像数据,然后分别在百元钞票的正、背面选取4个特征点,以分析百元真假钞票在正、背面的光谱反射率差异。从真假钞正面四个特征位置的光谱反射率曲线可知,不同特征点处,真钞与假钞之间的有些图案的光谱反射率差异较大,有些图案的光谱反射率则差异并不显著。而不同批次的假钞,其不同位置的光谱反射率也存在较大的差异。对于真假钞背面的分析可知,不同特征点,在不同的波段,真钞与假钞以及假钞之间的光谱反射率同样有一定的差异。根据真钞与假钞正、背面8个特征点的光谱反射率曲线变化特征,选取500,660和870 nm三个波长的灰度图,观察到真钞与假钞在不同特征点的灰度图均表现出明显的差异。真钞在500 nm处的图像轮廓清晰,在660和870 nm两个波长,无论是正面或背面,真钞均有多处特征位置有异于假钞,因此可用660或870 nm区别百元真假钞。为了突出真钞与假钞之间的图像差异,利用基于波段运算得到百元真钞与假钞的灰度图。由图可知,在正、背面上,真钞在多个地方均区别于假钞。由真假钞正面的前12个主成分的灰度图可知,无论从真、假钞正面或是背面,每一主成分均有真钞显著区别于假钞的地方。根据真假钞正背面的纹理特征图可知,真钞的纹理特征显著区别于假钞。研究表明百元真钞与不同版本的假钞在可见/近红外光谱范围内光谱反射率差异显著,近红外特征波段、光谱运算、主成分分析、纹理特征等技术均可探寻到百元真假钞在正、背面存在的位置差异。由此可知,可见/近红外高光谱成RMB is the legal currency of the people’s Republic of China.The authenticity of RMB is directly related to the harmony and stability of the society.As the most significant denomination bill in China,identifying the authenticity of 100 yuan RMB is more critical.In the paper,we use the visible/near-infrared hyperspectral imager to obtain the hyperspectral image data of an actual bill and two fake bills.Then select four feature points on the front and back of the bill to analyze the spectral differences between the front and back of the bill.From the spectral reflectance curves of the four characteristic positions on the front of the authentic and counterfeit banknote.It can be seen that the spectral reflectance of some patterns between the real and the counterfeit banknote varies greatly at different characteristic points,while that of some patterns is not significant.Moveover,for different batches of counterfeit banknotes,the spectral reflectance at different locations also has large differences.The analysis of the back of real and counterfeit banknotes shows that with different feature points,the spectral reflectivity between the real and the counterfeit banknote is also different in different bands.According to spectral reflectance curves of the 8 characteristic points on the front and back of the genuine and the counterfeit banknote,the grayscale images of the wavelengths at 500,660 and 870 nm were selected.It was observed that the grayscale images of real and counterfeit banknotes showed obvious differences at different feature points.The image contour of the real banknote at 500 nm is clear.In 660 and 870 nm,no matter on the front or back,the real banknote has several characteristic positions different from the counterfeit banknotes.Therefore,it can be used to distinguish the real 100-yuan banknote at 660 or 870 nm.In order to highlight the difference between real banknotes and counterfeit banknotes,the gray-scale image of real and counterfeit banknotes of 100 yuan is obtained by band calculation.As can be

关 键 词：高光谱成像 光谱分析 主成分分析 伪币鉴别 

分 类 号：S127[农业科学—农业基础科学]