Shape determination of unidimensional objects: the virtual image correlation method

Laboratoire FAST, Universite Paris Sud XI, 91405 Orsay

^a e-mail: marc.francois@u-psud.fr

Abstract

The proposed method, named Virtual Image Correlation, allows one to identify an analytical expression of the shape of a curvilinear object from its image. It uses a virtual beam, whose curvature field is expressed as a truncated mathematical series. The virtual beam width only needs to be close to the physical one; its gray level (in the transverse direction) is bell-shaped. The method consists in finding the coefficients of the series for which the correlation between physical and virtual beams is the best. The accuracy and the robustness of the method is shown by the mean of two examples. The first details a Young’s modulus identification from a cantilever beam image. The second is relative to a thermal plume image, that have a weak contrast and a lot of noise.