Résumé
Temperature variation induced by the deformation of a solid material is related to heat sources of various natures like energy dissipation or thermo-mechanical couplings. The heat sources are thus interesting data to understand and model the thermo-mechanical behavior of materials. Moreover the infrared thermography allows to measure the temperature fields on a specimen surface but the temperature fields depend on the thermal properties of the material, on the geometry and on the boundary conditions. It is thus necessary to identify heat sources from the temperature measurements. Various techniques exist in the literature but the difficulties mainly come from the noise of the experimental data. In this paper we propose to develop a preprocessing in order to reduce the noise using a decomposition of the measured temperature fields in a truncated basis made of proper orthogonal modes derived from a set of physically admissible thermal fields associated with the considered heat diffusion problem. After this preprocessing, classical methods like finite difference, spectral decomposition or local fitting, are applied and the results are compared.