Regensburg 2022 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

CPP: Fachverband Chemische Physik und Polymerphysik

CPP 7: Wetting, Fluidics and Liquids at Interfaces and Surfaces

CPP 7.3: Vortrag

Montag, 5. September 2022, 11:30–11:45, H39

Deep learning to analyze sliding drops — •Sajjad Shumaly, Fahimeh Darvish, Xiaomei Li, Alexander Saal, Chirag Hinduja, Werner Steffen, Oleksandra Kukharenko, Rüdiger Berger, and Hans-Jürgen Butt — Max Planck Institute for Polymer Research, Ackermannweg 10, D-55128, Mainz, Germany

Investigation of drop sliding forces requires knowledge of the shape of the drop close to the three-phase contact line. State-of-the-art contact angle measurement methods are designed for analyzing symmetric and high-resolution images of the drop. The analysis of videos of drops sliding down a tilted plane is hampered due to the low-resolution of the cutout area where the drop is visible. The drop is just a part of the whole image. In addition, drops sliding down a tilted plate are unsymmetrical in shape and the three-phase contact line may deform due to the sticky points.

In order to increase the accuracy of the measured contact angle, we trained a deep learning-based super-resolution model with an up-scale ratio of 3, i.e. the trained model is able to enlarge droplet images 9 times. In the second step, we performed an optimized polynomial fitting approach to measure the contact angle even for symmetric, asymmetric, or deformed droplets without the need for liquid parameters. To find the best parameters for polynomial fitting in our special problem, we conducted a systematic experiment using synthetic images.