Estimation of complex motion from thermographic image sequences

In this contribution a novel technique for computing complexmotion involving heat transport processes will be presented. Theproposed technique is a local gradient based approach, combiningtransport models with motion analysis. It allows for thesimultaneous estimation of both motion and parameter of anunderlying transport model. Since the analysis is based on thermalimage sequences, estimates are computed to a high temporal andspatial resolution, limited only by the resolution and frame rateof the employed IR camera. This novel technique was utilized onexchange processes at the atmosphere/ocean boundary, wheresignificant parameters of heat transfer could be measured and atransport model verified. Using the presented algorithms, surfaceflows as well as convergences and divergences on air-waterinterfaces can be measured accurately. Apart from applications inoceanography and botany, relevant benefits of the proposedtechnique to NDT will be presented. It is possible to compensatefor motion to reach accuracies much better than 1/10th of a pixel.Through the direct estimation of locally resolved diffusivities inmaterials, insights can be gained about defects present. Byestimating not only isotropic diffusion but also the whole matrixof anisotropic diffusion, the technique is highly relevant tomeasurements of composite materials.