Estimating Surface Orientation in Imaging Polarimetry of Solar Illuminated Outdoor Scenes

Abstract:

Observations from JPL’s Ground-based Multiangle SpectroPolarimetric Imager (GroundMSPI) are used to relate Angle of Linear Polarization (AoLP) and Degree of Linear Polarization (DoLP) measurements to the surface orientation of objects in solar illuminated outdoor scenes. Ground-MSPI is an eight-band spectropolarimetric camera mounted on a two-axis gimbal to acquire pushbroom imagery. The camera uses a photoelastic-modulator-based polarimetric imaging technique to measure linear Stokes parameters in three wavebands (470, 660, and 865 nm) with a 0.005 uncertainty in DoLP. A range of illumination geometries is acquired by collecting image data throughout the day; polarized light scattering models are evaluated over the range of sampled scattering angles.

In this work, we show how the orientation of a surface is related to the polarized light scattering from a solar illuminated object. We consider objects prevalent in outdoor scenery: grass, foliage, buildings, and roads; for these materials assuming a real-valued index of refraction is appropriate. The relationship between the orientation of reflected polarized light and the orientation of the surface heavily depends upon the texture of that surface relative to the wavelength of the light. For surfaces that are rough, relative to the wavelength, we observe a measured AoLP that is consistently perpendicular to the scattering plane. For smooth surfaces the AoLP measurements are projections of the surface normal vector onto the imaging plane (see Figure 1). The DoLP contains complementary information about surface orientation; the DoLP is related to the projection of the surface normal vector on to the scattering plane. Since the scattering plane changes with the sun position, increased temporal sampling throughout the day improves the reconstruction of this component of the surface normal vector. We demonstrate how to estimate the surface normal vector, for materials that are prevalent in outdoor scenery, by reconciling Ground-MSPI measurements with polarized light scattering models. We also report on the use of multiple wavebands and geometric shadowing functions to improve estimates of the surface normal vector. Implications for modeling the polarized light scattering of the Earth’s surface for top of atmosphere aerosol retrieval are included.