Polarization-based transparent surface modeling from two views

Abstract

In this paper, we propose a novel method to recover the surface shape of transparent objects.

Polarization-based inverse rendering from a single view

“…The angle of the polarizer where we observe I max will be the azimuth angle [18][19][20][21][22][23]. Since the linear polarizer has a cycle of 180 , we obtain two azimuth angles in the domain 0 < 360 .…”

“…Shape-from-polarization [18][19][20][21][22][23] estimate the surface normal of an object from a single view illuminated by incoherent light sources, even without knowing any illumination information such as intensity, color, shape, and direction. The polarization state of diffuse light as well as specular light depends on the surface normal of the object, thus, by analyzing the polarization state of diffuse light, we can obtain the information about the shape of the object.…”

A Geodesic Active Contour Framework for Finding Glass

“…On the other hand, transparent objects are characterized by the fact that one can see through them, rather than by any intrinsic texture pattern. The main focus of the machine vision research dealing with transparent objects has been concerned with 3D structure from X [4,9,15,16,22] and layer separation in the case of reflections [12,23]. Relatively little work has been concerned with the automatic identification of objects made of transparent materials [18].…”

A neural network scheme for transparent surface modelling

“…The first problem is addressed by introducing a whole-surface lighting (photometric sampler) [Saito et al 1999] or edge lighting (multiple parallel linear lighting) for rotationally symmetric object. While the second problem is addressed by introducing additional information such as thermal radiation [Miyazaki et al 2002], new view image [Miyazaki et al 2003], trinocular stereo [Rahmann and Canterakis 2001], or object's symmetry ].…”