Research On The Method Of Spectral Analysis Of Printed Matter With Visible-Short Wavelength Infrared Spectrum Data

Spectral analysis is a fast,efficient and nondestructive detection technique,the subtle differences in the composition of matter can be expressed by the spectral information.Therefore,This article combines spectral analysis with print inspection,and study the possibility of using spectral analysis to effectively distinguish between prints using different inks.so as to explore a more appropriate method for spectral analysis of printed matter.In this paper,the ASD portable spectrometer was used to collect the spectra of fries boxes and packets.The spectral data of different samples of the two types of prints showed different degrees of difference,among them the difference between the spectra of the different samples of the fries box is the smaller,and the differences of the cigarette packets are larger.Before the analysis of the spectral data,the original spectral data need to be processed by spectral data smoothing,multiple scattering correction and envelope removing.In this paper,two general spectral matching methods are adopted.The first one is based on spectral absorption feature matching,including absorption peak distribution matching and absorption peak feature parameter matching.In this paper,an automatic spectral absorption peak extraction algorithm based on equal interval average method is proposed,which can achieve different spectral absorption peaks.The experimental results show that the matching method of absorption peak distribution and the matching of absorption peak characteristic parameters can achieve effective discrimination of spectral data,method of feature parameter matching has high accuracy,and it's accuracy rate reaches 88.9~100.0%.The second spectral analysis method is the overall spectral similarity matching,in this paper,three spectral similarity measurement parameters are selected,such as correlation coefficient,spectral angle,and Jaccard similarity coefficient.Three methods are used to match the printed spectral data,matching space Visible light,near infrared,visible-short wavelength infrared three spectral space.In addition,three methods were also used to match the spectral data of printed products in visible,short wavelength infrared and visible-near-infrared spectral spaces.The results show that for the three types of print data,the correlation coefficients and the spectral angle spectral matching accuracy are higher than the Jaccard similarity coefficient matching.The combined spectral matching method,the correlation coefficient-spectral angle combination of the matching method has the highest accuracy.In general,for the analysis of printed spectral data,the preferred spectral matching method is the spectral matching method of absorption peak position distribution matching,correlation coefficient matching,spectral angle matching,and correlation coefficient-spectral angle combination,and the spectral analysis technique can be used analysis of differences in printed spectrum.