Particle image reconstruction for particle detection in particle tracking velocimetry

Abstract

A new methodology for particle identification and localization in the context of particle tracking velocimetry (PTV) is presented. The aim is to overcome the issue of inherent detection errors under high particle density conditions. The approach is based on the particle position reconstruction through the inversion of a linear model connecting the PTV signal with a particle-based representation of the 3D-to-2D projection. The inversion procedure accounts for both the non-negativity and the sparsity of the sought solution. Simulation tests using synthetically generated images are carried out to evaluate the sensitivity of the proposed method to characteristic parameters such as the particle image density, the particle image size, the model image size, and/or background noise. Its ability to provide better detection performances with higher reliability than conventional techniques is demonstrated.