The Interference Imaging Spectrometer (IIM), as one of the most important payloads on Chang'E-1 satellite, has acquired large amount of hyperspectral data for retrieving element and mineral information of the lunar surface. It's still a great challenge to retrieve mineral distribution with IIM data, due to the fact that the spectral range of IIM data is limited to 480-960nm which does not covers the main lunar minerals' absorption centers. At present, there is few precise quantitative mineral distribution of the lunar surface to be used for the validation of the retrieved results. In this study, we implement a linear simulation of IIM data to analyze the accuracy of mineral endmember extraction with IIM data. In the linear simulation, Gaussian response model is used to resample the lunar mineral sample spectra, which are measured by RELAB of Brown University, to the same spectral center and resolution with IIM data, and random noises of the same trend with IIM data are added into the simulated data. The lunar mare and highland are simulated according to the different composition of plagioclase, clinopyroxene, olivine and ilmenite. Four endmember extraction approaches (VCA/ICA/MVSA/ SISAL) are employed to extract endmembers and spectral angle distance (SAD) is used as criterion for accuracy analysis. The results show that all the SADs of endmembers extracted by MVSA and SISAL are lower than 0.1, and that of MVSA is a little lower than SISAL. Plagioclase, of which the SADs are lower than 0.015, can always be extracted by all approaches. All SADs of pyroxene and ilmenite are lower than 0.1. Olivine, of which the SADs are high to 0.35 when the ICA approach is implemented, has the lowest extraction accuracy. This may be caused by the unobvious V-NIR spectral features of olivine spectrum.
[1]
José M. Bioucas-Dias,et al.
A variable splitting augmented Lagrangian approach to linear spectral unmixing
,
2009,
2009 First Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing.
[2]
Erkki Oja,et al.
Independent component analysis: algorithms and applications
,
2000,
Neural Networks.
[3]
José M. Bioucas-Dias,et al.
Vertex component analysis: a fast algorithm to unmix hyperspectral data
,
2005,
IEEE Transactions on Geoscience and Remote Sensing.
[4]
L. Qiao,et al.
Estimation of lunar titanium content: Based on absorption features of Chang’E-1 interference imaging spectrometer (IIM)
,
2010
.
[5]
Paul G. Lucey,et al.
Mineral maps of the Moon
,
2003
.
[6]
Xia Zhang,et al.
Global absorption center map of the mafic minerals on the Moon as viewed by CE-1 IIM data
,
2010
.