| Summary: | Approved for public release; distribution is unlimited === Visible to near-, shortwave-, and longwave-infrared (VNIR, SWIR, LWIR) remote sensing data are typically analyzed in their individual wavelength regions, even though theory suggests combined use would emphasize complementary features. This research explored the potential for improvements in material classification using integrated datasets. Hyperspectral (HSI) VNIR and SWIR data from the MaRSuper Sensor System (MSS-1) were analyzed with HSI LWIR data from the Spatially Enhanced Broadband Array Spectrograph System (SEBASS) to determine differences between individual (baseline) and combined analyses. The first integration approach applied separate minimum noise fraction (MNF) transforms to the three regions and combined only non-noise transformed bands from the individual regions during analysis. The second approach integrated over 470 hyperspectral bands covering the VNIR, SWIR, and LWIR wavelengths before using MNF analysis to isolate linear band combinations containing high signal to noise. Spectral endmembers isolated from data were unmixed using partial unmixing. The feasible and high abundance pixels were spatially mapped using a consistent feasibility ratio threshold. Both integration methods enabled straight-forward and effective identification, characterization, and mapping of the scene because higher variability existed between endmembers and background. Results were compared to the baseline analysis. Material identification was more conclusive when analyzing across the full spectrum.
|
|---|
