期刊
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS
卷 117, 期 -, 页码 61-69出版社
ELSEVIER SCIENCE BV
DOI: 10.1016/j.chemolab.2012.02.002
关键词
Orthogonal projection; Pretreatment; Preprocessing; Subspace; Linear model
This article describes several linear pretreatments based on orthogonal projections. The main differences of these pretreatments lie in the way the information to be removed are identified, using calibration dataset, pure spectra, experimental designs or mathematical models. Removing all the undesired spectral information yields spectra proportional to the net analyte signal, so it is important to collect the most complete information possible, using the complementarities of different approaches. The correction should then be processed with a single Euclidian orthogonal projection that gathers all the information, rather than with successive operations. By embedding Euclidian orthogonal projections into the calibration, it is not necessary to reapply them to new datasets. (C) 2012 Elsevier B.V. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据