PUMA
Istituto di Scienza e Tecnologie dell'Informazione     
Pavoni G., Dellepiane M., Callieri M., Scopigno R. Automatic selection of video frames for path regularization and 3D reconstruction. In: GCH 2016 - Eurographics Workshop on Graphics and Cultural Heritage (Genova, Italy, 5-7 October 2016). Proceedings, article n. 1. Eurographics, 2016.
 
 
Abstract
(English)
Video sequences can be a valuable source to document the state of objects and sites. They are easy to acquire and they usually ensure a complete coverage of the object of interest. One of their possible uses is to recover the acquisition path, or the 3D shape of the scene. This can be done by applying structure-from-motion techniques to a representative set of frames extracted from the video. This paper presents an automatic method for the extraction of a predefined number of representative frames that ensures an accurate reconstruction of the sequence path, and possibly enhances the 3D reconstruction of the scene. The automatic extraction is obtained by analyzing adjacent frames in a starting subset, and adding/removing frames so that the distance between them remains constant. This ensures the reconstruction of a regularized path and an optimized coverage of all the scene. Finally, more frames are added in the portions of the sequence when more detailed objects are framed. This ensures a better description of the sequence, and a more accurate dense reconstruction. The method is automatic, fast and independent from any assumption about the acquired object or the acquisition strategy. It was tested on a variety of different video sequences, showing that a satisfying result can be obtained regardless of the length and quality of the input.
URL: http://https://diglib.eg.org/handle/10.2312/gch20161376
DOI: 10.2312/gch.20161376
Subject Video analysis
Frame extraction
3D reconstruction
I.3.7 COMPUTER GRAPHICS. Three-Dimensional Graphics and Realism
I.4.1 IMAGE PROCESSING AND COMPUTER VISION. Digitization and Image Capture
I.2.10 ARTIFICIAL INTELLIGENCE. Vision and Scene Understanding


Icona documento 1) Download Document PDF


Icona documento Open access Icona documento Restricted Icona documento Private

 


Per ulteriori informazioni, contattare: Librarian http://puma.isti.cnr.it

Valid HTML 4.0 Transitional