Thermographic flow visualization by means of non-negative matrix factorization

In order to investigate the areas of different flow regimes in the boundary layer of an airfoil, thermography is a powerful flow visualization tool. However, the distinguishability between boundary layer flow regimes such as laminar or turbulent is limited due to systematic and random inhomogeneity in the measured temperature field, hindering a clear separation of the flow regimes. In order to increase the distinguishability of different flow regimes, a time series of thermographic images is evaluated by means of a non-negative matrix factorization. As a result, the non-negative matrix factorization creates images that contain the dominant features of the measured images, while reducing systematic temperature gradients within the flow regimes by up to a factor of five. This way an increase of the distinguishability between every pair of consecutive flow regimes can be achieved on the surface of a non-heated cylinder in cross-flow condition. As a further application example of the non-negative matrix factorization, the distinguishability between the flow and the laminar-turbulent transition zone on a heated helicopter airfoil is also increased by a factor of five. Hence, non-negative matrix factorization is capable of enhancing thermographic flow visualization for increasing the distinguishability of different flow phenomena.