Skip to Main content Skip to Navigation
Master Thesis

Filtrage de mesures de trajectoires de véhicules

Abstract : Drivers individual behavior on a microscopic scale stay relatively unknown. Existing models lack corroboration data and cannot be completely trusted. In order of providing proof of accuracy, datasets based on traffic video recording were collected. These video datasets are heavily impacted by measurement errors. Recording missing parts, camera resolution or post-processing treatment (mainly vehicle identification between successive frames) are most of the error’s sources. Literature proposes a wide variety of solution from classical data treatment such as Kalman or Butterworth filters to adding extra physical constraints. In 2016, Buisson, Villegas and Rivoirard suggested a new methodology based on polar coordinates and tested it with good results on the MOCoPo dataset. The present work continues this way and applies it on four others dataset with again good results provided that the trajectories stay in hypothesises constraints (which are relatively low: temporal continuity, no circuit ).
Document type :
Master Thesis
Complete list of metadata
Contributor : Bibliothèque Entpe Connect in order to contact the contributor
Submitted on : Tuesday, May 25, 2021 - 12:44:52 PM
Last modification on : Tuesday, July 6, 2021 - 3:31:27 AM
Long-term archiving on: : Thursday, August 26, 2021 - 7:24:13 PM


Files produced by the author(s)




  • HAL Id : dumas-03234468, version 1



Sébastien Plantier. Filtrage de mesures de trajectoires de véhicules. Sciences de l'ingénieur [physics]. 2020. ⟨dumas-03234468⟩



Les métriques sont temporairement indisponibles