Comparison of spectral selection methods in the development of classification models from visible near infrared hyperspectral imaging data
Applications of hyperspectral imaging (HSI) to the quantitative and qualitative measurement of samples have grown widely in recent years, due mainly to the improved performance and lower cost of imaging spectroscopy instrumentation. Data sampling is a crucial yet often overlooked step in hyperspectr...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IM Publications Open
2019-01-01
|
Series: | Journal of Spectral Imaging |
Subjects: | |
Online Access: | https://www.impopen.com/download.php?code=I08_a4 |
id |
doaj-b6fbb05334424046adc1cf7bff256f85 |
---|---|
record_format |
Article |
spelling |
doaj-b6fbb05334424046adc1cf7bff256f852020-11-25T02:08:49ZengIM Publications OpenJournal of Spectral Imaging2040-45652040-45652019-01-0181a410.1255/jsi.2019.a4Comparison of spectral selection methods in the development of classification models from visible near infrared hyperspectral imaging dataAoife A. Gowen0Jun-Li Xu1Ana Herrero-Langreo2UCD School of Biosystems and Food Engineering, University College of Dublin (UCD), Belfield, Dublin 4, IrelandUCD School of Biosystems and Food Engineering, University College of Dublin (UCD), Belfield, Dublin 4, IrelandUCD School of Biosystems and Food Engineering, University College of Dublin (UCD), Belfield, Dublin 4, IrelandApplications of hyperspectral imaging (HSI) to the quantitative and qualitative measurement of samples have grown widely in recent years, due mainly to the improved performance and lower cost of imaging spectroscopy instrumentation. Data sampling is a crucial yet often overlooked step in hyperspectral image analysis, which impacts the subsequent results and their interpretation. In the selection of pixel spectra for the calibration of classification models, the spatial information in HSI data can be exploited. In this paper, a variety of sampling strategies for selection of pixel spectra are presented, exemplified through five case studies. The strategies are compared in terms of the proportion of global variability captured, practicality and predictive model performance. The use of variographic analysis as a guide to the spatial segmentation prior to sampling leads to the selection of representative subsets while reducing the variation in model performance parameters over repeated random selection.https://www.impopen.com/download.php?code=I08_a4hyperspectral imagingdata samplingclassificationspatialvariographic analysis |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Aoife A. Gowen Jun-Li Xu Ana Herrero-Langreo |
spellingShingle |
Aoife A. Gowen Jun-Li Xu Ana Herrero-Langreo Comparison of spectral selection methods in the development of classification models from visible near infrared hyperspectral imaging data Journal of Spectral Imaging hyperspectral imaging data sampling classification spatial variographic analysis |
author_facet |
Aoife A. Gowen Jun-Li Xu Ana Herrero-Langreo |
author_sort |
Aoife A. Gowen |
title |
Comparison of spectral selection methods in the development of classification models from visible near infrared hyperspectral imaging data |
title_short |
Comparison of spectral selection methods in the development of classification models from visible near infrared hyperspectral imaging data |
title_full |
Comparison of spectral selection methods in the development of classification models from visible near infrared hyperspectral imaging data |
title_fullStr |
Comparison of spectral selection methods in the development of classification models from visible near infrared hyperspectral imaging data |
title_full_unstemmed |
Comparison of spectral selection methods in the development of classification models from visible near infrared hyperspectral imaging data |
title_sort |
comparison of spectral selection methods in the development of classification models from visible near infrared hyperspectral imaging data |
publisher |
IM Publications Open |
series |
Journal of Spectral Imaging |
issn |
2040-4565 2040-4565 |
publishDate |
2019-01-01 |
description |
Applications of hyperspectral imaging (HSI) to the quantitative and qualitative measurement of samples have grown widely in recent years, due mainly to the improved performance and lower cost of imaging spectroscopy instrumentation. Data sampling is a crucial yet often overlooked step in hyperspectral image analysis, which impacts the subsequent results and their interpretation. In the selection of pixel spectra for the calibration of classification models, the spatial information in HSI data can be exploited. In this paper, a variety of sampling strategies for selection of pixel spectra are presented, exemplified through five case studies. The strategies are compared in terms of the proportion of global variability captured, practicality and predictive model performance. The use of variographic analysis as a guide to the spatial segmentation prior to sampling leads to the selection of representative subsets while reducing the variation in model performance parameters over repeated random selection. |
topic |
hyperspectral imaging data sampling classification spatial variographic analysis |
url |
https://www.impopen.com/download.php?code=I08_a4 |
work_keys_str_mv |
AT aoifeagowen comparisonofspectralselectionmethodsinthedevelopmentofclassificationmodelsfromvisiblenearinfraredhyperspectralimagingdata AT junlixu comparisonofspectralselectionmethodsinthedevelopmentofclassificationmodelsfromvisiblenearinfraredhyperspectralimagingdata AT anaherrerolangreo comparisonofspectralselectionmethodsinthedevelopmentofclassificationmodelsfromvisiblenearinfraredhyperspectralimagingdata |
_version_ |
1724925153212628992 |