Optothermal depth profiling by neural network infrared radiometry signal recognition

The feasibility of a neural network radiometric photothermal depth profiling method is verified using well-defined artificial samples with varying optical properties across the layers. The signal calculation model is shown to be accurate and the neural network approach to solve the inverse problem is shown to be feasible. Both from simulated and experimental radiometric signals, accurate reconstructions are obtained for heat source and optical-absorption coefficient profiles.