Shadow-insensitive Material Detection/classification with Atmospherically Corrected Hyperspectral Imagery

Hyperspectral imaging (HSI) from airborne or space-based platforms, currently conducted mainly in the 0.4 2.5 μm wavelength region, is a valuable technique for detection and classification of materials and objects on the Earth's surface. In a typical analysis, the spectral radiance data are "corrected" or compensated for the atmospheric and illumination conditions to yield spectral reflectance, and the results are processed with any of a variety of algorithms, which may, for example, compare the contents of each pixel with reference spectra for classification, search the scene for a desired material spectrum, or calculate a terrain property such as a vegetation cover index. Atmospheric correction algorithms include several based on first-principles radiation transport models (e.g., Gao et al., 1996; Green et al., 1996; Richter, 1996; Matthew et al., 2000) as well as the Empirical Line Method (ELM), which utilizes known materials in the scene. An alternative approach is to work with the original radiance data; this is effective for identifying scene anomalies and for supervised processing, where the scene elements can be identified visually by an analyst.