par Bonatto, Daniele
;Dury, Sarah
;Sancho Aragon, Jaime;Juarez, Eduardo;Lafruit, Gauthier
;Teratani, Mehrdad 
Référence Proceedings (International Multimedia Modelling Conference), 15523 LNCS, page (128-140)
Publication Publié, 2025-10-01




Référence Proceedings (International Multimedia Modelling Conference), 15523 LNCS, page (128-140)
Publication Publié, 2025-10-01
Article révisé par les pairs
Résumé : | Plenoptic cameras capture 3D content by utilizing micro-parallax through a micro-lens array placed in front of their sensor. While this design eliminates the need for stereo cameras or separate depth sensors to extract 3D information, certain scenarios -such as handling disocclusions, capturing non-Lambertian surfaces, or imaging large scenes from multiple perspectives- require a multi-camera setup. Capturing a scene with multiple plenoptic cameras introduces new challenges due to the complex camera design and the lack of tailored image-processing methods for the plenoptic format. In this paper, we present the first comprehensive pipeline for acquiring and processing video sequences using multiple plenoptic cameras of different types. Our approach addresses synchronization, color correction and cameras’ parameters calibration. We validate our method on a new video dataset featuring both a Raytrix R8 and Raytrix R32 cameras. |