摘要
利用小波变换对５２个烟草样品的近红外漫反射光谱进行滤波处理，并用ＰＬＳ法来计算烟草样品的总氮含量，结果表明小波变换滤波后，预测集的相对标准偏差由原来的９．２％降为７．４％，此结果也优于傅里叶交换和五点三次平滑。
The noise of NIR diffusive reflectance spectra of 52 kinds of tobacco is filtered by using wavelet transform. The total nitrogen content in tobacco samples is calculated by PLS method. The result of signal smoothing is satisfactory. The RSD of prediction set is reduced to 7. 4% from 9. 2% of original spectra. The result of wavelet transform filtering is superior to multiple smoothing (a five-points span) and Fourier transform filtering.
出处
《分析化学》
SCIE
EI
CAS
CSCD
北大核心
1998年第1期34-37,共4页
Chinese Journal of Analytical Chemistry
关键词
小波变换
近红外
滤波
烟草分析
总氮
Wavelet transform, near-infrared, noise filtering, tobacco analysis
