Title :
Multibaseline stereo using a single-lens camera
Author :
Amtoun, Mohamed ; Boufama, Boubakeur
Author_Institution :
Sch. of Comput. Sci., Windsor Univ., Ont., Canada
Abstract :
In stereopsis, correspondence is a central problem that is known to suffer from several inherent difficulties. Single-lens stereo systems have brought a geometrical solution to some of the problems of stereopsis. The plenoptic camera is one of those systems that we found to simplify the process of depth recovery, but whose main limitation lies in its small baseline. For this particular reason, matches in the plenoptic camera are obtained more easily, however, this trades off against the accuracy with which they are recovered. In this paper, we show how its lack of accuracy due to a small baseline can be made up for to some extent, by using a multibaseline estimate in lieu of the commonly used equal-baseline estimate.
Keywords :
photographic lenses; stereo image processing; depth recovery; disparity estimation; multibaseline stereo; plenoptic camera; single-lens camera; stereopsis problem; Cameras; Computer science; Frequency; Geometry; Layout; Lenses; Optical design; Solid modeling; Stereo vision; Yield estimation;
Conference_Titel :
Image Processing, 2003. ICIP 2003. Proceedings. 2003 International Conference on
Print_ISBN :
0-7803-7750-8
DOI :
10.1109/ICIP.2003.1246983