DocumentCode :
1790636
Title :
Surface normal vector force driven 3D object reconstruction via Graph-cut
Author :
Sujung Kim ; Seong Dae Kim
Author_Institution :
Dept. of Electr. Eng., KAIST, Daejeon, South Korea
fYear :
2014
fDate :
22-25 June 2014
Firstpage :
1
Lastpage :
2
Abstract :
Multi-view 3D surface reconstruction is typically formulated as minimizing an appropriate energy functional, which tends to find the empty set, i.e. an undesirable result, especially when cooperating only with a boundary term. So generally a regional term is employed, which indicates inside or outside of the surface. It is, however, the challenge because inherently there is no straightforward way to assign interior or exterior of the surface since photo-consistency on the surface boundary is the main source of geometric information. In this paper, we address this problem by proposing a surface normal vector force driven regional term which is a data-aware ballooning term and cast it into a Graph-cut framework. In experiments, we verify the superiority of the proposed algorithm on various multiple view data sets.
Keywords :
computer vision; graph theory; image reconstruction; vectors; boundary term; data-aware ballooning term; energy functional; geometric information; graph-cut framework; multiple view data sets; multiview 3D surface reconstruction; photo-consistency; regional term; surface normal vector force; Force; Image reconstruction; Kernel; Reliability; Surface reconstruction; Three-dimensional displays; Vectors; 3D object reconstruction; data-aware ballooning term; graphcut framework; multi-view stereo; surface orientation;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Consumer Electronics (ISCE 2014), The 18th IEEE International Symposium on
Conference_Location :
JeJu Island
Type :
conf
DOI :
10.1109/ISCE.2014.6884544
Filename :
6884544
Link To Document :
بازگشت