High Resolution Particle Detection via Spectral Estimation

Within this paper the problem of Particle Tracking Velocimetry (PTV) is considered. PTV investigates turbulences in fluids by tracking particles with high temporal and spatial resolution. Due to the acquisition with high speed cameras, high spatial resolution and, therefore, a huge amount of data for the captured pictures, the particle flow can only be observed within a very small period of time. The proposed, new and practical system design allows a significant decrease of the spatial sampling rate for the pictures. This decreases the amount of data to be captured and, therefore, increases the possible time for observation. Designing a 2-stage sampling allows the use of conventional camera sensors as used up to now. Thus, it is directly applicable without a change of existing camera systems. From a signal processing point of view, this paper considers the problem of sampling a 2-D signal with Finite Rate of Innovation (FRI). Therefore, an appropriate solution for sampling and reconstruction of such signals is proposed. Furthermore, the performance for reconstruction of signal parameters under the influence of coloured noise is investigated as well as the accuracy of the estimates if no model knowledge is available.