Shape Estimation of Transparent Objects by Using Polarization Analyses

Abstract

Techniques developed in the field of computer graphics and virtual reality are applied in many situations today, with the result that methods of measuring three-dimensional shapes of real objects are seen to be more and more important. However, few methods are proposed to measure three-dimensional shapes of transparent objects such as those made of glass and acrylic. In this thesis, I propose three methods for estimating the surface shapes of transparent objects by using polarization analysis. There is an ambiguity problem that results from using polarization data obtained by ob-serving reflected light from one view. The first method resolves the ambiguity problem by