3D Reconstruction by polarimetric imaging method based on perspective model

Classical 3D inspection systems require users to coat transparent objects before measurement. Experimental techniques via non contact measurement, suggested in literature, do not treat inter refections. The aim of our work is to develop a non contact 3D measurement system for transparent objects by using a polarimetric imaging method in far infrared range. The classical approach relies on the use of orthographic model generated by a telecentric lens in practical setup. However telecentric lenses working in far infrared range are not available. Therefore, we have to adapt pinhole model corresponding to non-telecentric lenses for shape from polarization. In this paper we introduce a 3D reconstruction method to exploit polarimetric imaging with perspective model. We also propose two mathematical approaches in order to reduce reconstruction error : data analysis method to better estimate Stokes parameters and a validation method after Stokes parameters estimation. These techniques are applicable irrespective of the nature of the selected model and
any linear system resolution.

SPIE Europe Optical Metrology, Munich, Germany, 15 June 2009.

Rindra RANTOSON, Christophe STOLZ, David FOFI, Fabrice MERIAUDEAU

Reconstruction 3D par Polarimétrie basée sur le Modèle Perspectif

GDR ISIS – Journée Imagerie Optique Non Conventionnelle, Paris, France, 11 Mars 2009.

Rindra RANTOSON, Christophe STOLZ, David FOFI, Fabrice MERIAUDEAU