Learning fluid trajectory models for time-resolved PIV
| Autor: | Godet, Pierre, Champagnat, Frédéric, Le Besnerais, Guy, Plyer, Aurélien |
|---|---|
| Přispěvatelé: | DTIS, ONERA, Université Paris Saclay (COmUE) [Palaiseau], ONERA-Université Paris Saclay (COmUE), André, Cécile |
| Jazyk: | angličtina |
| Rok vydání: | 2019 |
| Předmět: | |
| Zdroj: | ISPIV 2019 ISPIV 2019, Jul 2019, MUNICH, Germany |
| Popis: | International audience; We present a new multiframe cross-correlation (Lucas-Kanade based) algorithm for time-resolved PIV. This algorithm leverages time coherence in image sequences by decomposing the temporal dependency of motion on an arbitrarily chosen trajectory basis. We propose to learn this basis from the data by performing a Principal Component Analysis on trajectories sampled from the studied sequence. We show on simulated data that such an approach can outperform the polynomial models classically used in multiframe PIV. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|