Joint polarization and event sensing for surface normal reconstruction

Surface normal reconstruction from combining linear polarization ​and standard monochrome cameras has been widely studied in the past years. The aim of this work is to study for the first time the possibility of reconstructing the surface normals of a scene from the data collected by exploiting a linear polarizing filter and an event sensing camera, where the latter is a novel sensor whose applications are still an open field in the computer vision community. For this task a deep learning algorithm has been used in order to perform the conversion of the event camera's signal as if it was acquired from a standard monochrome camera, then the surface normals are reconstructed with an already well known procedure.