CNNs trained on GAN-generated, auto-labeled data classify defocused particle images as tracers versus bubbles or droplets with 95-100% accuracy on most tested datasets.
Interferometric Particle Imaging for Particle Sizing in the Front-, Side-, and Back-Scatter Region
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
Interferometric particle imaging is a widely used optical measuring technique for the sizing of poly-dispersed spherical particles like droplets and bubbles. In its conventional approach, the method is limited to forward-scattering angles and therefore, requiring a second optical access, restricting the range of possible applications. In the present work, this limitation of the scattering angle is addressed, showing that also other scattering angles, especially in the back-scatter region are applicable, expanding the technique to applications with only a single optical access. A general method for the identification of suitable scattering angles both for droplets and bubbles is proposed. The visibility criterion for interference patterns from particles is generalized and possible glare point parings and their separation in the forward-, side- and back-scatter regimes are discussed for droplets and bubbles. Due to being the most popular examples, different scattering angles are proposed for water droplets and air bubbles in water. In the last part, the method is validated on a bubble sizing experiment.
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Pattern-Based Phase-Separation of Tracer and Dispersed Phase Particles in Two-Phase Defocusing Particle Tracking Velocimetry
CNNs trained on GAN-generated, auto-labeled data classify defocused particle images as tracers versus bubbles or droplets with 95-100% accuracy on most tested datasets.