Research on Hyperspectral Unmixing Oil Spill Monitoring

DOI

10.2991/icsd-16.2017.40How to use a DOI?

Keywords

remote sensing; hyperspectral unmixing; oil spill; endmember; abundance estimation

Abstract

Hyperspectral unmixing is an important technique for hyperspectral image analysis. In this paper, we took Airborne Visible Infrared Imaging Spectrometer (AVIRIS) hyperspectral imagery as dataset to monitor oil spills. The information of oil spills was retrieved through image preprocessing, minimum noise fraction (MNF) feature extraction, endmember extraction (pure pixel index (PPI), unsupervised orthogonal subspace projection (UOSP)) and fully constrained least squares (FCLS) abundance estimation. In the steps of endmember extraction, the experiment measured endmember spectra of oil and water were used as reference spectra. Then we compared the endmember spectra extracted in the image to the measured spectra by the spectral angle. At last the FCLS abundance estimation was carried on to evaluate the endmember extraction quality. The result demonstrates that the unsupervised OSP-FCLS model is better than supervised PPI-FCLS endmember extraction.

Copyright

© 2017, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Can Cui
AU  - Ying Li
AU  - Hong-Ji Chen
AU  - Bing-Xin Liu
AU  - Jin Xu
AU  - Guan-Nan Li
PY  - 2016/12
DA  - 2016/12
TI  - Research on Hyperspectral Unmixing Oil Spill Monitoring
BT  - Proceedings of the 2nd 2016 International Conference on Sustainable Development (ICSD 2016)
PB  - Atlantis Press
SP  - 185
EP  - 189
SN  - 2352-5401
UR  - https://doi.org/10.2991/icsd-16.2017.40
DO  - 10.2991/icsd-16.2017.40
ID  - Cui2016/12
ER  -