In-situ 1-kHz real-time particle tracking velocimetry using high-speed streaming camera

This paper describes the use of a high-speed streaming camera combined with in-house image processing for real-time particle tracking velocimetry (PTV) above 1 kHz. The proposed system is evaluated by measuring the diameter and velocity of multiple free-falling droplets in a simultaneous manner. Two existing particle tracking methods are examined: median flow (MF) and four-frame best estimate (4BE). Area-limiting processing is newly added to MF and 4BE to reduce the region of interest (ROI) in which particles are detected in the next image, resulting in area-limiting MF (AMF) and area-limiting 4BE (A4BE). Except for MF, the rest three methods of AMF, 4BE, and A4BE are capable of 1-kHz real-time PTV for 1246 pixel x 600 pixel images, with 4BE and A4BE achieving even at 10 kHz for 1246 pixel x 100 pixel images. The processing times for tracking and image preparation are measured, and the bottleneck is found to be the image acquisition and preparation, rather than the tracking. AMF achieves faster processing times than MF in all conditions, showing the effectiveness of the limited ROI, while 4BE and A4BE exhibit comparable performance. The measurement error is confirmed to be approximately 1% for droplet velocity and diameter, demonstrating the high accuracy of the first in-situ real-time PTV exceeding 1 kHz.