Polarimetric Remote Sensing in the Visible to Near Infrared

Polarization adds another dimension to the spatial intensity and spectral information typically acquired in remote sensing. Polarization imparted by surface reflections contains unique and discriminatory signatures which may augment spectral target-detection techniques. Benefits such as improving man-made object detection are often touted, as well as possible improvements to spectral algorithms used for detection and identification. However, virtually all efforts fail to cast polarimetric remote sensing within a cohesive framework in which a priori predictions of polarized radiance are made, as is done with spectral remote sensing techniques. This is due, in part, to the challenges of measuring and representing polarization signatures. This research develops the governing radiometric equation for polarimetric remote sensing, illustrating the role of the polarimetric bidirectional reflectance distribution function (pBRDF). Using the governing radiometric equation as a basis, a technique for quantifying the pBRDF of background materials is presented. The measurement technique enables the generation of pBRDF as a function of the ground separation distance (GSD). Empirical data from employing the technique are fit to a pBRDF model. The model enables extrapolation of results to arbitrary illumination and viewing conditions throughout the visible to near infrared, all as a function of GSD. A target material pBRDF model is also developed which enables the pBRDF representation of spatially homogeneous surfaces, typical of man-made materials. This model uses an unpolarized BRDF model as the basis, and generalizes the representation while enabling polarization. BRDF model parameters are determined from an existing database of BRDF measurements, which enables “polarization” of the database. The quantification and understanding of target and background material polarization signatures is a prerequisite for exploring methods of fusing polarimetric and spectral information. This work advances this understanding and lays the foundation for spectralpolarimetric target detection techniques and algorithm development.