Skip to Main content Skip to Navigation
Conference papers

An Omni-RGB+D Camera Rig Calibration and Fusion using Unified Camera Model for 3D Reconstruction

Abstract : The perfect vision system could be a system which can obtain surround images or information at once. We present a vision system which can view the image in 360 with high-resolution depth information. The proposed vision system is compact and rigid with two fisheye cameras that provide a 360-degree field of view(FoV). Alongside, a high-resolution stereo vision camera is mounted to monitor anterior FoV for precise depth perception of the scene. To effectively calibrate the proposed camera system, we offer a novel camera calibration approach taking the advantages of Unified Camera Model representation. The proposed calibration method outperforms the state-of-the-art methods. Moreover, we proposed more affective algorithm in fusing the two fisheye images into a single unified sphere, which offers seamless stitching results. This new omni-vision rig system is designed to obtain sufficient information to be used on a robot for object detection and recognition. A large scale SLAM and dense 3D reconstruction can be achieved taking the advantage of the large FoV.
Document type :
Conference papers
Complete list of metadatas

Cited literature [14 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01525780
Contributor : Ahmad Zawawi Jamaluddin <>
Submitted on : Monday, May 22, 2017 - 11:35:36 AM
Last modification on : Friday, July 17, 2020 - 2:54:11 PM
Document(s) archivé(s) le : Wednesday, August 23, 2017 - 2:26:27 PM

File

zawawi_qcav2017_18052017.pdf
Files produced by the author(s)

Identifiers

Citation

Ahmad Jamaluddin, Osama Mazhar, Cansen Jiang, Ralph Seulin, Olivier Morel, et al.. An Omni-RGB+D Camera Rig Calibration and Fusion using Unified Camera Model for 3D Reconstruction. 13th International Conference on Quality Control by Artificial Vision 2017, May 2017, Tokyo, Japan. ⟨10.1117/12.2266945⟩. ⟨hal-01525780⟩

Share

Metrics

Record views

455

Files downloads

860