A novel event-based ensemble particle tracking velocimetry for single-pixel turbulence statistics

Abstract

This paper presents a long-duration event-based ensemble particle tracking velocimetry (EEPTV) method for single-pixel turbulence statistics; toward this end, a turbulent annular jet at a Reynolds number of 7,500 is used for demonstration. Leveraging the low-data-redundancy, high-temporal-resolution capabilities of an event-based camera, the EEPTV system complemented with grayscale correction from a low-speed frame-based camera successfully recovers 2.8 × 10⁶ image frames of the flow at 2,000 Hz. The EEPTV accurately captures the high velocity gradient in the shear layers, and the mean absolute velocity discrepancy is only 0.09 pixels/frame. This method also demonstrates superior performance in resolving Reynolds stresses compared to conventional window-based PIV, which suffers from an underestimation of up to 52 % due to the spatial smoothing effect. A detailed analysis of sampling errors using the bootstrap resampling method reveals that the widths of statistical confidence intervals follow a power-law relationship with respect to both the frame count and the bin size, highlighting the necessity of long-duration acquisition for accurate single-pixel turbulence measurements. In this way, an efficient framework for high-resolution turbulence statistics is established, overcoming the long-standing trade-off between the spatial resolution and statistical fidelity. This work has the potential to provide reliable and spatially resolved turbulence statistics for turbulence modelling and data-driven algorithms.