TY - GEN
T1 - Towards robust vegetation indices
T2 - 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014
AU - Aasen, Helge
AU - Gnyp, Martin Leon
AU - Miao, Yuxin
AU - Bareth, Georg
N1 - Publisher Copyright:
© 2014 IEEE.
Copyright:
Copyright 2019 Elsevier B.V., All rights reserved.
PY - 2014/6/28
Y1 - 2014/6/28
N2 - Hyperspectral vegetation indices (HVIs) have shown great potential for characterizing and monitoring vegetation and agricultural crops. Additionally, hyperspectral data becomes more commonly available. Latter may be used to address varying annual crop growth. In this paper we describe the multi-correlation matrix strategy as a new approach to derive robust HVIs from multiple hyperspectral field spectrometers datasets. The approach combines the information from multiple correlation matrices (CMs). The software HyperCor is used to automate the data pre-processing and CMs computation. In this study we use data from three growth stages (tillering, stem elongation, heading) in five years (2007-2009, 2011 and 2012) to estimate rice biomass. The new approach is validated through leave-one-out cross-validation and compared to results from a direct approach. On average the multi-correlation matrix approach showed 15% better performance and could reduce the RMSE compared to the direct approach.
AB - Hyperspectral vegetation indices (HVIs) have shown great potential for characterizing and monitoring vegetation and agricultural crops. Additionally, hyperspectral data becomes more commonly available. Latter may be used to address varying annual crop growth. In this paper we describe the multi-correlation matrix strategy as a new approach to derive robust HVIs from multiple hyperspectral field spectrometers datasets. The approach combines the information from multiple correlation matrices (CMs). The software HyperCor is used to automate the data pre-processing and CMs computation. In this study we use data from three growth stages (tillering, stem elongation, heading) in five years (2007-2009, 2011 and 2012) to estimate rice biomass. The new approach is validated through leave-one-out cross-validation and compared to results from a direct approach. On average the multi-correlation matrix approach showed 15% better performance and could reduce the RMSE compared to the direct approach.
KW - HyperCor
KW - biomass
KW - hyperspectral data processing
KW - multi-correlation matrix strategy
KW - rice
KW - robust vegetation indices
UR - http://www.scopus.com/inward/record.url?scp=85038565741&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85038565741&partnerID=8YFLogxK
U2 - 10.1109/WHISPERS.2014.8077547
DO - 10.1109/WHISPERS.2014.8077547
M3 - Conference contribution
AN - SCOPUS:85038565741
T3 - Workshop on Hyperspectral Image and Signal Processing, Evolution in Remote Sensing
BT - 2014 6th Workshop on Hyperspectral Image and Signal Processing
PB - IEEE Computer Society
Y2 - 24 June 2014 through 27 June 2014
ER -