Format

Send to

Choose Destination
Sensors (Basel). 2017 Aug 11;17(8). pii: E1857. doi: 10.3390/s17081857.

Translational Imaging Spectroscopy for Proximal Sensing.

Author information

1
Section 1.4 Remote Sensing, Helmholtz Centre Potsdam-GFZ German Research Centre for Geosciences, Telegrafenberg, 14473 Potsdam, Germany. christian.rogass@gfz-potsdam.de.
2
Section 1.4 Remote Sensing, Helmholtz Centre Potsdam-GFZ German Research Centre for Geosciences, Telegrafenberg, 14473 Potsdam, Germany. friederike.koerting@gfz-potsdam.de.
3
Section 1.4 Remote Sensing, Helmholtz Centre Potsdam-GFZ German Research Centre for Geosciences, Telegrafenberg, 14473 Potsdam, Germany. christian.mielke@gfz-potsdam.de.
4
Section 1.4 Remote Sensing, Helmholtz Centre Potsdam-GFZ German Research Centre for Geosciences, Telegrafenberg, 14473 Potsdam, Germany. maximilian.brell@gfz-potsdam.de.
5
Section 1.4 Remote Sensing, Helmholtz Centre Potsdam-GFZ German Research Centre for Geosciences, Telegrafenberg, 14473 Potsdam, Germany. nina.boesche@gfz-potsdam.de.
6
Section 1.4 Remote Sensing, Helmholtz Centre Potsdam-GFZ German Research Centre for Geosciences, Telegrafenberg, 14473 Potsdam, Germany. maria.bade@gfz-potsdam.de.
7
Section 1.4 Remote Sensing, Helmholtz Centre Potsdam-GFZ German Research Centre for Geosciences, Telegrafenberg, 14473 Potsdam, Germany. christian.hohmann@gfz-potsdam.de.

Abstract

Proximal sensing as the near field counterpart of remote sensing offers a broad variety of applications. Imaging spectroscopy in general and translational laboratory imaging spectroscopy in particular can be utilized for a variety of different research topics. Geoscientific applications require a precise pre-processing of hyperspectral data cubes to retrieve at-surface reflectance in order to conduct spectral feature-based comparison of unknown sample spectra to known library spectra. A new pre-processing chain called GeoMAP-Trans for at-surface reflectance retrieval is proposed here as an analogue to other algorithms published by the team of authors. It consists of a radiometric, a geometric and a spectral module. Each module consists of several processing steps that are described in detail. The processing chain was adapted to the broadly used HySPEX VNIR/SWIR imaging spectrometer system and tested using geological mineral samples. The performance was subjectively and objectively evaluated using standard artificial image quality metrics and comparative measurements of mineral and Lambertian diffuser standards with standard field and laboratory spectrometers. The proposed algorithm provides highly qualitative results, offers broad applicability through its generic design and might be the first one of its kind to be published. A high radiometric accuracy is achieved by the incorporation of the Reduction of Miscalibration Effects (ROME) framework. The geometric accuracy is higher than 1 μpixel. The critical spectral accuracy was relatively estimated by comparing spectra of standard field spectrometers to those from HySPEX for a Lambertian diffuser. The achieved spectral accuracy is better than 0.02% for the full spectrum and better than 98% for the absorption features. It was empirically shown that point and imaging spectrometers provide different results for non-Lambertian samples due to their different sensing principles, adjacency scattering impacts on the signal and anisotropic surface reflection properties.

KEYWORDS:

geology; geometry; hyperspectral; imaging spectroscopy; laboratory; processing chain; reflectance

Supplemental Content

Full text links

Icon for Multidisciplinary Digital Publishing Institute (MDPI) Icon for PubMed Central
Loading ...
Support Center