Multi-Camera System Calibration for 3D Scene Reconstruction, Modeling and Visualization
| Autor: | Mijatov, Jovan, Zlokolica, Vladimir, Obradović, Ratko, Ljubobratović, Relja, Obradović, Radovan, Mastilović, Vladimir, Vdovjak, Krešimir, Galić, Irena |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2018 |
| Předmět: |
Computer Science::Computer Vision and Pattern Recognition
camera calibration computer vision photogrammetry 3D modeling multi-view stereo structure-from-motion architecture design technology fabrication education ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Astrophysics::Instrumentation and Methods for Astrophysics |
| Popis: | Current state-of-the-art multi-camera-view 3D reconstruction algorithms provide high quality 3D point cloud, which is further used for 3D meshing, texture mapping and finally generating corresponding 3D model of the scene. One of the most important factors for both 3D surface reconstruction and texture mapping is accurate and robust camera calibration. This paper presents a novel approach for multicamera system calibration where we extend and make mixture of known calibration techniques to calibrate multi- camera system with specially designed three- dimensional calibration object. More precisely, we extend calibration technique that uses planar pattern to three-dimensional pattern camera calibration. First, camera parameters are roughly estimated for all cameras in multiple camera system, then these parameters are refined using standard bundle adjustment algorithm. The proposed approach has been tested with synthetic and real data. Mean square error between re- projected and detected pattern points has been found to be in the range of 0.1px – 0.5px, for real camera data of 24MPx. The proposed method is shown to be more accurate or comparable to other calibration techniques. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|