3D digitization of transparent objects by polalization techniques in IR & by triangulation in UV

Two non-conventional methods for the 3D digitization of transparent objects via non-contact measurement are reported in this thesis. 3D digitization is a well acknowledged technique for opaque objects and various commercial solutions based on different measurement approaches are available in the market offering different types of resolution at different prices. Since these techniques require a diffused or lambertian surface, their application to transparent surfaces fails. Indeed, rays reflected by the transparent surface are perturbed by diverse inter-reflections induced by the refractive properties of the object. Therefore, in industrial applications like quality control, the transparent objects are powder coated followed by their digitization. However, this method is expensive and can also produce inaccuracies. Among the rare methods suggested in the literature, shape from polarization provides reliable results even though their accuracy had to be improved by coping with the inter-reflections. The two proposed solutions handle the extension of the existing methods to wavelengths beyond visible ranges: - shape from polarization in Infra Red (IR) range to deal with the above-mentioned inter-reflections; - scanning by Ultra Violet (UV) laser (based on triangulation scheme) to overcome the refraction problem that can be feasibly applied in industrial applications. The characteristic physical properties of transparent objects led us to explore the IR and UV ranges; since, transparent glass has strong absorption bands in the IR and UV ranges and therefore has opaque appearance. The first approach exploits the specular reflection of the considered object surface in IR and the second one exploits the fluorescence property of the object when irradiated with UV rays. Shape from polarization traditionally based on telecentric lenses had to be adapted with non-telecentric lenses to be used in the IR range. Thus, an approximation of the orthographic model is developed in this thesis while a validation method is implemented and integrated in the reconstruction process after Stokes parameters estimation, in order to improve the accuracy of the results. Some results of digitized objects are presented, which prove the feasibility of the shape from polarization method in the IR range to be used for transparent objects. A total of four configurations of the triangulation system are implemented in this thesis to exploit fluorescence produced by the UV laser scanning of the second approach. Experimental investigations aimed at characterizing the fluorescence are done. A specific fluorescence tracking method is carried out to deal with the inherent noise in the acquisitions. The uniqueness of the method relies on the criteria that are derived from the analysis of spectroscopic results. A validation method is made to optimize the configuration system while reducing the accuracy of reconstruction error. The results of some object digitization are presented with accuracies better than previously reported works.

Data and Resources

Additional Info

Field Value
Source https://theses.hal.science/tel-00692460
Author Sanders, Rindra, Rantoson
Maintainer CCSD
Last Updated May 20, 2026, 10:54 (UTC)
Created May 20, 2026, 10:54 (UTC)
Identifier NNT: 2011DIJOS039
Language fr
Rights https://about.hal.science/hal-authorisation-v1/
contributor Laboratoire Electronique, Informatique et Image [UMR6306] (Le2i) ; Université de Bourgogne (UB)-École Nationale Supérieure d'Arts et Métiers (ENSAM)-AgroSup Dijon - Institut National Supérieur des Sciences Agronomiques, de l'Alimentation et de l'Environnement-Centre National de la Recherche Scientifique (CNRS)
creator Sanders, Rindra, Rantoson
date 2011-11-03T00:00:00
harvest_object_id e6a0bca3-edeb-4bdd-bb49-e5eb8d3ba9ab
harvest_source_id 3374d638-d20b-4672-ba96-a23232d55657
harvest_source_title test moissonnage SELUNE
metadata_modified 2026-03-31T00:00:00
set_spec type:THESE