On the generation of reduced models by Proper Orthogonal Decomposition from experimental image data

This paper illustrates an image processing technique using the Proper Orthogonal Decomposition (POD) of infrared thermal data for the construction of reduced models starting from experimental data. We consider a thin steel plate with a point heat source in the middle activated at time=0. A TELOPS DSP-83 fast IR camera collects a sequence of infrared image data. The 2D samples constitute the data set used to generate a POD empirical basis. Galërkin projection of the heat conduction PDE onto the basis generates the finite dimensional approximate dynamical system. Results are compared with experimental data and analytical solution.