Multiband Spectropolarimetric Analysis for Man-Made Material Delineation

Traditional multispectral remote sensing methods that rely on nonpolarized light reflectance signatures for material identification often have difficulty in delineating similar materials within broad classes, such as vegetation, soil, and man-made materials, because the spectral signatures are similar. However, since polarization is influenced by feature shape and roughness, polarized reflectance signatures may be able to separate materials that fall within these broad classes. This study investigates how material surface metrics relate to the spectropolarimetric responses of man-made materials in laboratory experiments and aerial imagery. This paper provides background on feature detection using polarimetic data, the laboratory experiments that have been conducted, and an overview of analysis methods.