Assessment and application of optical flow in background-oriented schlieren for compressible flows

Optical flow provides an opportunity to elevate the resolution and sensitivity of deflection sensing in background-oriented schlieren (BOS). Despite extensive relevant literature within the field of computer vision, there is a lack of proper quantification of its abilities and limitations with regard to the state-of-the-art BOS experiments. Thus, this study performs an assessment of accuracy and resolution limits in different flow field scenarios utilizing background patterns generated with random dot and wavelet noise distributions. Accordingly, a synthetic assessment over a theoretically generated Prandtl–Meyer expansion fan is conducted with variations introduced in the background patterns and operational parameters of optical flow. A clear superiority of accuracy and resolvable range of density gradient amplitudes over cross-correlation is demonstrated. Moreover, an experimental assessment of supersonic flow features over multiple wind tunnel models is performed. The influence of experimental constraints, limitations and uncertainties related to the application of optical flow in BOS and its comparative performance against the block-matching counterpart is characterized. Graphical abstract