Multi-scale Direct Sparse Visual Odometry for Large-Scale Natural Environment
International audience In this paper, we describe a multi-scale monocular direct sparse visual odometry (DSO) system to recover large-scale trajectories in unstructured natural environments in real time, while building a consistent metric map of the visited scenes. In contrast to the current state-o...
Main Authors: | , |
---|---|
Other Authors: | , , |
Format: | Conference Object |
Language: | English |
Published: |
HAL CCSD
2018
|
Subjects: | |
Online Access: | https://hal.archives-ouvertes.fr/hal-02278006 |
id |
ftccsdartic:oai:HAL:hal-02278006v1 |
---|---|
record_format |
openpolar |
spelling |
ftccsdartic:oai:HAL:hal-02278006v1 2023-05-15T16:00:47+02:00 Multi-scale Direct Sparse Visual Odometry for Large-Scale Natural Environment Wu, Xiaolong Pradalier, Cédric Georgia Tech Lorraine Metz Université de Franche-Comté (UFC) Université Bourgogne Franche-Comté COMUE (UBFC)-Université Bourgogne Franche-Comté COMUE (UBFC)-Ecole Supérieure d'Electricité - SUPELEC (FRANCE)-Georgia Institute of Technology Atlanta -CentraleSupélec-Ecole Nationale Supérieure des Arts et Metiers Metz-Centre National de la Recherche Scientifique (CNRS) Verona, France 2018-09-05 https://hal.archives-ouvertes.fr/hal-02278006 en eng HAL CCSD IEEE hal-02278006 https://hal.archives-ouvertes.fr/hal-02278006 2018 International Conference on 3D Vision (3DV) https://hal.archives-ouvertes.fr/hal-02278006 2018 International Conference on 3D Vision (3DV), Sep 2018, Verona, France. pp.89-97 Image reconstruction Cameras Optimization Three-dimensional displays Real-time systems Tracking Simultaneous localization and mapping [INFO]Computer Science [cs] info:eu-repo/semantics/conferenceObject Conference papers 2018 ftccsdartic 2021-11-07T01:41:09Z International audience In this paper, we describe a multi-scale monocular direct sparse visual odometry (DSO) system to recover large-scale trajectories in unstructured natural environments in real time, while building a consistent metric map of the visited scenes. In contrast to the current state-of-the-art DSO system, the proposed method allows for more robust motion estimation and more accurate reconstruction in distant scenes by exploiting the characteristics of short- and long-range pixels, respectively. The long-range pixels, which are less sensitive to small camera translations, are used to initialize the camera rotation, so as to boost the tracking robustness in challenging natural environments. A multi-scale reconstruction framework is developed to recover short-range structure over successive frames, as well as the long-range structure over distant frames, hence allowing for a more consistent mapping precision. The reconstruction precision, the tracking accuracy, and the robustness of the proposed system are extensively evaluated with a publicly available vKITTI dataset, as well as the challenging Devon Island dataset, and Symphony Lake dataset. A detailed performance comparison between the proposed method and the state-of-the-art DSO system is presented. Conference Object Devon Island Archive ouverte HAL (Hyper Article en Ligne, CCSD - Centre pour la Communication Scientifique Directe) Devon Island ENVELOPE(-88.000,-88.000,75.252,75.252) |
institution |
Open Polar |
collection |
Archive ouverte HAL (Hyper Article en Ligne, CCSD - Centre pour la Communication Scientifique Directe) |
op_collection_id |
ftccsdartic |
language |
English |
topic |
Image reconstruction Cameras Optimization Three-dimensional displays Real-time systems Tracking Simultaneous localization and mapping [INFO]Computer Science [cs] |
spellingShingle |
Image reconstruction Cameras Optimization Three-dimensional displays Real-time systems Tracking Simultaneous localization and mapping [INFO]Computer Science [cs] Wu, Xiaolong Pradalier, Cédric Multi-scale Direct Sparse Visual Odometry for Large-Scale Natural Environment |
topic_facet |
Image reconstruction Cameras Optimization Three-dimensional displays Real-time systems Tracking Simultaneous localization and mapping [INFO]Computer Science [cs] |
description |
International audience In this paper, we describe a multi-scale monocular direct sparse visual odometry (DSO) system to recover large-scale trajectories in unstructured natural environments in real time, while building a consistent metric map of the visited scenes. In contrast to the current state-of-the-art DSO system, the proposed method allows for more robust motion estimation and more accurate reconstruction in distant scenes by exploiting the characteristics of short- and long-range pixels, respectively. The long-range pixels, which are less sensitive to small camera translations, are used to initialize the camera rotation, so as to boost the tracking robustness in challenging natural environments. A multi-scale reconstruction framework is developed to recover short-range structure over successive frames, as well as the long-range structure over distant frames, hence allowing for a more consistent mapping precision. The reconstruction precision, the tracking accuracy, and the robustness of the proposed system are extensively evaluated with a publicly available vKITTI dataset, as well as the challenging Devon Island dataset, and Symphony Lake dataset. A detailed performance comparison between the proposed method and the state-of-the-art DSO system is presented. |
author2 |
Georgia Tech Lorraine Metz Université de Franche-Comté (UFC) Université Bourgogne Franche-Comté COMUE (UBFC)-Université Bourgogne Franche-Comté COMUE (UBFC)-Ecole Supérieure d'Electricité - SUPELEC (FRANCE)-Georgia Institute of Technology Atlanta -CentraleSupélec-Ecole Nationale Supérieure des Arts et Metiers Metz-Centre National de la Recherche Scientifique (CNRS) |
format |
Conference Object |
author |
Wu, Xiaolong Pradalier, Cédric |
author_facet |
Wu, Xiaolong Pradalier, Cédric |
author_sort |
Wu, Xiaolong |
title |
Multi-scale Direct Sparse Visual Odometry for Large-Scale Natural Environment |
title_short |
Multi-scale Direct Sparse Visual Odometry for Large-Scale Natural Environment |
title_full |
Multi-scale Direct Sparse Visual Odometry for Large-Scale Natural Environment |
title_fullStr |
Multi-scale Direct Sparse Visual Odometry for Large-Scale Natural Environment |
title_full_unstemmed |
Multi-scale Direct Sparse Visual Odometry for Large-Scale Natural Environment |
title_sort |
multi-scale direct sparse visual odometry for large-scale natural environment |
publisher |
HAL CCSD |
publishDate |
2018 |
url |
https://hal.archives-ouvertes.fr/hal-02278006 |
op_coverage |
Verona, France |
long_lat |
ENVELOPE(-88.000,-88.000,75.252,75.252) |
geographic |
Devon Island |
geographic_facet |
Devon Island |
genre |
Devon Island |
genre_facet |
Devon Island |
op_source |
2018 International Conference on 3D Vision (3DV) https://hal.archives-ouvertes.fr/hal-02278006 2018 International Conference on 3D Vision (3DV), Sep 2018, Verona, France. pp.89-97 |
op_relation |
hal-02278006 https://hal.archives-ouvertes.fr/hal-02278006 |
_version_ |
1766396800340590592 |