Naslov
Multi-Camera System Calibration for 3D Surround Camera Distribution

Autori
Mijatov, Jovan ; Zlokolica, Vladimir ; Ljubobratović, Relja ; Vdovjak, Krešimir ; Galić, Irena ; Obradović, Ratko V. Mastilović

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni

Izvornik
Fifth Croatian Computer Vision Workshop - Abstract Book / Lončarić, Sven ; Cupec, Robert - Osijek : University of Zagreb, Croatia ; Josip Juraj Strossmayer University of Osijek, Croatia, 2016, 18-18

Skup
5th Croatian Computer Vision Workshop (CCVW 2016)

Mjesto i datum
Osijek, Hrvatska, 11.10.2016

Vrsta sudjelovanja
Poster

Vrsta recenzije
Domaća recenzija

Ključne riječi
multi-camera calibration ; 3D reconstruction ; dense point cloud
(kalibracija kamere ; 3D rekonstrukcija ; gusti oblak točaka)

Sažetak
Multi-camera system with 3D surround camera configuration is currently most common setup used for high precision 3D reconstruction and modeling in computer graphics applications. Within this camera setup the most important features are exact camera positioning and camera characteristics which must be optimally tuned for 3D object, i.e., scene that is being reconstructed. One of the most important steps in 3D reconstruction is multi-camera calibration, which should be optimized for target application ; 3D calibration object and methodology is to be appropriately designed. In this paper we present new approach for multi-camera system calibration for 3D surround camera configuration, aimed at 3D reconstruction and 3D modeling. The proposed calibration approach estimates intrinsic parameters and lens distortions of each camera as well as extrinsic parameters that define relative 3D positioning of the cameras within the multi-camera system. The proposed 3D object used for calibration is specifically designed 3D cube, which is first designed and tested in virtual environment to meet all constraints set by target application ; our application is 3D face reconstruction and modeling. The 3D calibration cube is formed of different patterns which can be viewed by more than one camera at a time allowing thus proposed sequential stereo calibration approach. The proposed calibration approach is evaluated by comparing estimated multi-camera system parameters with two state of the art calibration approaches and based on quality of the dense 3D point cloud generation.

Izvorni jezik
Engleski

Znanstvena područja
Računarstvo

POVEZANOST RADA