Snapshot hyperspectral imaging probe with principal component analysis and confidence ellipse for classification
Hyperspectral imaging combines imaging and spectroscopy to provide detailed spectral information for each spatial point in the image. This gives a three-dimensional spatial-spatial-spectral datacube with hundreds of spectral images. Probe-based hyperspectral imaging systems have been developed so th...
Saved in:
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Article |
Language: | English |
Published: |
2019
|
Subjects: | |
Online Access: | https://hdl.handle.net/10356/106362 http://hdl.handle.net/10220/49580 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Nanyang Technological University |
Language: | English |
id |
sg-ntu-dr.10356-106362 |
---|---|
record_format |
dspace |
spelling |
sg-ntu-dr.10356-1063622023-03-04T17:22:22Z Snapshot hyperspectral imaging probe with principal component analysis and confidence ellipse for classification Lim, Hoong-Ta Murukeshan, Vadakke Matham Asundi, Anand K. School of Mechanical and Aerospace Engineering Fifth International Conference on Optical and Photonics Engineering Centre for Optical and Laser Engineering Engineering::Mechanical engineering Hyperspectral Imaging Fiber Bundle Hyperspectral imaging combines imaging and spectroscopy to provide detailed spectral information for each spatial point in the image. This gives a three-dimensional spatial-spatial-spectral datacube with hundreds of spectral images. Probe-based hyperspectral imaging systems have been developed so that they can be used in regions where conventional table-top platforms would find it difficult to access. A fiber bundle, which is made up of specially-arranged optical fibers, has recently been developed and integrated with a spectrograph-based hyperspectral imager. This forms a snapshot hyperspectral imaging probe, which is able to form a datacube using the information from each scan. Compared to the other configurations, which require sequential scanning to form a datacube, the snapshot configuration is preferred in real-time applications where motion artifacts and pixel misregistration can be minimized. Principal component analysis is a dimension-reducing technique that can be applied in hyperspectral imaging to convert the spectral information into uncorrelated variables known as principal components. A confidence ellipse can be used to define the region of each class in the principal component feature space and for classification. This paper demonstrates the use of the snapshot hyperspectral imaging probe to acquire data from samples of different colors. The spectral library of each sample was acquired and then analyzed using principal component analysis. Confidence ellipse was then applied to the principal components of each sample and used as the classification criteria. The results show that the applied analysis can be used to perform classification of the spectral data acquired using the snapshot hyperspectral imaging probe. MOE (Min. of Education, S’pore) Published version 2019-08-07T06:53:13Z 2019-12-06T22:09:53Z 2019-08-07T06:53:13Z 2019-12-06T22:09:53Z 2017 Journal Article Lim, H.-T., & Vadakke Matham, M. (2017). Snapshot hyperspectral imaging probe with principal component analysis and confidence ellipse for classification. Proceedings of SPIE - Fifth International Conference on Optical and Photonics Engineering, 10449, 104491O-. doi:10.1117/12.2270736 0277-786X https://hdl.handle.net/10356/106362 http://hdl.handle.net/10220/49580 10.1117/12.2270736 en Proceedings of SPIE - Fifth International Conference on Optical and Photonics Engineering © 2017 SPIE. All rights reserved. This paper was published in Proceedings of SPIE - Fifth International Conference on Optical and Photonics Engineering and is made available with permission of SPIE. 6 p. application/pdf |
institution |
Nanyang Technological University |
building |
NTU Library |
continent |
Asia |
country |
Singapore Singapore |
content_provider |
NTU Library |
collection |
DR-NTU |
language |
English |
topic |
Engineering::Mechanical engineering Hyperspectral Imaging Fiber Bundle |
spellingShingle |
Engineering::Mechanical engineering Hyperspectral Imaging Fiber Bundle Lim, Hoong-Ta Murukeshan, Vadakke Matham Snapshot hyperspectral imaging probe with principal component analysis and confidence ellipse for classification |
description |
Hyperspectral imaging combines imaging and spectroscopy to provide detailed spectral information for each spatial point in the image. This gives a three-dimensional spatial-spatial-spectral datacube with hundreds of spectral images. Probe-based hyperspectral imaging systems have been developed so that they can be used in regions where conventional table-top platforms would find it difficult to access. A fiber bundle, which is made up of specially-arranged optical fibers, has recently been developed and integrated with a spectrograph-based hyperspectral imager. This forms a snapshot hyperspectral imaging probe, which is able to form a datacube using the information from each scan. Compared to the other configurations, which require sequential scanning to form a datacube, the snapshot configuration is preferred in real-time applications where motion artifacts and pixel misregistration can be minimized. Principal component analysis is a dimension-reducing technique that can be applied in hyperspectral imaging to convert the spectral information into uncorrelated variables known as principal components. A confidence ellipse can be used to define the region of each class in the principal component feature space and for classification. This paper demonstrates the use of the snapshot hyperspectral imaging probe to acquire data from samples of different colors. The spectral library of each sample was acquired and then analyzed using principal component analysis. Confidence ellipse was then applied to the principal components of each sample and used as the classification criteria. The results show that the applied analysis can be used to perform classification of the spectral data acquired using the snapshot hyperspectral imaging probe. |
author2 |
Asundi, Anand K. |
author_facet |
Asundi, Anand K. Lim, Hoong-Ta Murukeshan, Vadakke Matham |
format |
Article |
author |
Lim, Hoong-Ta Murukeshan, Vadakke Matham |
author_sort |
Lim, Hoong-Ta |
title |
Snapshot hyperspectral imaging probe with principal component analysis and confidence ellipse for classification |
title_short |
Snapshot hyperspectral imaging probe with principal component analysis and confidence ellipse for classification |
title_full |
Snapshot hyperspectral imaging probe with principal component analysis and confidence ellipse for classification |
title_fullStr |
Snapshot hyperspectral imaging probe with principal component analysis and confidence ellipse for classification |
title_full_unstemmed |
Snapshot hyperspectral imaging probe with principal component analysis and confidence ellipse for classification |
title_sort |
snapshot hyperspectral imaging probe with principal component analysis and confidence ellipse for classification |
publishDate |
2019 |
url |
https://hdl.handle.net/10356/106362 http://hdl.handle.net/10220/49580 |
_version_ |
1759854147023142912 |