Shape Estimation of Transparent Objects by Using Inverse Polarization Ray Tracing
IEEE Transactions on Pattern Analysis and Machine Intelligence
International Journal of Computer Vision
Hi-index | 0.00 |
A new method for multiple viewpoint 3D shape reconstruction is presented that relies on the polarization properties of surface reflection. The method is intended to complement existing stereo techniques by establishing correspondence for surfaces without salient features. The phase and degree of polarization from two views of an object are used to reconstruct surface patches. Local surface properties are then used to both align these patches and to compute a cost for that alignment. This cost is used as a basis to establish correspondence between the two views. The method is tested on an object library comprising shapes of varying complexity and material. An accuracy assessment is also presented where real world data are compared to ground truth.