A comprehensive polarisation model for surface orientation recovery

In this paper a polarisation model which predicts surface reflection as a function of refractive index and angle of incidence is introduced. We present the underlying physics of polarisation which is based on the Fresnel theory and Malus' law. The proposed model can be used to recover the shape of the objects for images taken under polarised light. The traditional way of shape recovery using diffuse polarisation is inaccurate due to noise in the degree of polarisation (DOP) measurements, computed from images captured under unpolarised light, which are small at most locations on the surface. The proposed model improves results as the DOP values are relatively larger in polarised light, we present a number of experimental results to demonstrate.