Joint estimation of objects geometry and material labels from single view polarized hyperspectral imagery

This paper presents a novel method for extracting 3D information and material labels from a scene by means of a hyperspectral polarimetric passive imaging sensor. This sensor provides full Stokes vector at each sensor pixel location, from which, degree, angle of linear polarization and indices of refraction are computed. Angle of linear polarization provides the azimuth angle of the surface normal vector. The depression angle of this surface normal vector is obtained in terms of the emitting object's spectral index of refraction, from the solution of a set of equations derived from Fresnel equations, snell's law, and Sellmeier dispersion formula. Results of the application of this approach on simulated spectral infrared polarimetric data are provided.