Development of stereoscopic three-dimensional particle tracking velocimetry using machine learning technique

A three-dimensional Particle Tracking Velocimetry (3D-PTV) technique is a method to extract velocity data inside target fluid. Unlike a three-dimensional Particle Image Velocimetry (3D-PIV) technique, the 3D-PTV technique focuses on three-dimensional movement of particles inside the target. Thus, it is appropriate to analyze sedimentation patterns during droplet evaporation phenomena. In this study, the 3D-PTV technique was applied to visualize evaporation flow pattern inside a binary mixture droplet. The technique was developed by a machine learning technique and verified by a numerical simulation. Then, the three-dimensional flow structure inside the droplet was obtained. The droplet consists of DI water, ethanol, and particles with a diameter of 15 µm. The images of particles inside the droplet were captured by two high-speed cameras. The images were revised by the image processing procedure, and successfully used for reconstructing the three-dimensional trajectories of the particles inside the droplet.