共 50 条
On the geometry of SNV and MSC
被引:194
|作者:
Fearn, Tom
[1
]
Riccioli, Cecilia
[2
]
Garrido-Varo, Ana
[2
]
Guerrero-Ginel, Jose Emilio
[2
]
机构:
[1] UCL, Dept Stat Sci, London WC1E 6BT, England
[2] Univ Cordoba, Dept Anim Prod, Cordoba 14014, Spain
关键词:
Circle;
Ellipse;
Outlier;
Scores;
Multiplicative scatter correction;
MSC;
Standard normal variate;
SNV;
Imaging spectroscopy;
D O I:
10.1016/j.chemolab.2008.11.006
中图分类号:
TP [自动化技术、计算机技术];
学科分类号:
0812 ;
摘要:
The spectral pre-treatments known as standard normal variate (SNV) and multiplicative scatter correction (MSC) often give very similar results, and are widely regarded as exchangeable. However their geometry in spectral space is not the same, and the results of this are sometimes apparent. SNV has the capacity to induce curved structures in score plots derived from the treated spectra, whilst MSC has a tendency to produce outliers in these plots. These phenomena are illustrated with an example from a near infrared imaging camera, and explained geometrically. (C) 2008 Elsevier B.V. All rights reserved.
引用
收藏
页码:22 / 26
页数:5
相关论文