Shake The Box’ - a 4D PTV algorithm: Accurate and ghostless reconstruction of Lagrangian tracks in densely seeded flows

The ‘Shake The Box’ (STB) PTV - method for the accurate reconstruction of particle tracks in densely seeded flows was first introduced in the scope of the 10th International Symposium on Particle Image Velocimetry (PIV13) in 2013 [Schanz 2013b]. In this first work, it was shown that the proposed method of incorporating the temporal information into the process of particle position reconstruction is able to improve results in various aspects compared to conventionally used techniques. However, only an experimental dataset was available to explore the abilities of the technique [Schanz 2013b]. In order to quantify the key features of the STB-method - and to identify the benefits and drawbacks – a scheme for the creation of synthetic tracks from input - vector volumes was created, allowing the calculation of physically (sufficiently) correct track systems at arbitrary seeding densities and time separations. Projections of such track systems on virtual cameras are used as input data for both STB, as well as an MLOS-SMART [Atkinson 2009] tomographic reconstruction algorithm. The results are compared with respect to positional accuracy of reconstructed particles, occurrence of ghost particles and reconstruction time. It is shown that, after a sufficient number of snapshots to allow for convergence of the algorithm, the STB-method is able to correctly identify nearly all particles in existence, even for high seeding density up to 0,125 particles per pixel (ppp). On the same time, the average position error stays very low (0,023 px for 0.125 ppp) – a gain of more than an order of magnitude compared to MLOS-SMART. The occurrence of ghost particles is almost completely suppressed (ghost level below 0.2 percent for 0.125 ppp). The efficient utilization of the temporal information allows a significant reduction in processing time, with STB being 4 to 6 times faster compared to MLOS-SMART. As the temporal information plays such a dominant role in the STB method – both in the gained results, as well as in the reconstruction process itself – the method cannot be described as pure particle reconstruction, but rather as a method to reconstruct the full temporal development of tracer particles within their residence in the measurement volume. Therefore it is more appropriate to speak of a track reconstruction scheme - or 4D PTV