Real-time Vehicle Motion Estimation Using Texture Learning and Monocular Vision

Abstract : High integrity localization system is an important challenge to improve safety for road vehicles. A way to meet the requirements is to fuse information from several sensors, from position and orientation sensors to motion, speed and acceleration sensors. This paper tackles the problem of vehicle motion estimation using monocular vision. A geometric model of the road is used to learn a texture patch in the current image, this patch is then tracked through the successive frames to estimate in real time the motion of the vehicle. The proposed method was assessed using a centimeter accuracy Real Time Kinematic GPS receiver.
Document type :
Conference papers
Liste complète des métadonnées

Cited literature [13 references]  Display  Hide  Download

https://hal.inria.fr/inria-00117116
Contributor : Yann Dumortier <>
Submitted on : Thursday, November 30, 2006 - 12:51:44 PM
Last modification on : Friday, May 25, 2018 - 12:02:03 PM
Document(s) archivé(s) le : Tuesday, April 6, 2010 - 7:35:52 PM

File

ICCVG06.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : inria-00117116, version 1

Collections

Citation

Yann Dumortier, Rodrigo Benenson, Mikael Kais. Real-time Vehicle Motion Estimation Using Texture Learning and Monocular Vision. ICCVG, Oct 2006, Varsovie. ⟨inria-00117116⟩

Share

Metrics

Record views

975

Files downloads

479