Interest in polarization for physics-based modeling, computer graphics, and vision algorithms has
grown since commercial off-the-shelf (COTS) polarimetric cameras first hit the marketplace in 2018.
The ability to track polarization in physics-based rendering engines and evaluate the polarization sen-
sitivity of capture systems has increased in relevance since polarimetric measurements became widely
accessible. Prior studies have confirmed that polarimetry offers unique information about a scene;
however, computational demands remain a barrier to exploring new applications using physics-based
representations of polarized light–matter interactions. COTS polarimeters will continue to proliferate
in the foreseeable future, and the development of computational methods for efficient forward synthesis
and inverse analysis will lay the foundations for optimal information extraction that maximizes their
application space.
Mueller matrices map incoming to outgoing partially-polarized states. Given a Mueller measure-
ment and a model of the polarized bidirectional reflectance distribution function (pBRDF) of an object,
an inverse problem estimates object attributes. Since polarized light scattering depend on illumina-
tion, wavelength, material, and shape of an object many inverse problems are plausible. In particular,
the inverse problem of shape and depth from-polarization, has applications in computer vision and
graphics. Solving these inverse problems requires the pBRDF to be modeled and represented both
accurately and efficiently.