DocumentCode :
650567
Title :
Computation of virtual environment from stereo-panoramic view
Author :
Audu, A.I. ; Sadka, A.H.
Author_Institution :
Electron. & Comput. Eng. Dept., Brunel Univ., Uxbridge, UK
fYear :
2013
fDate :
7-8 Oct. 2013
Firstpage :
1
Lastpage :
4
Abstract :
Depth map texturing is one of the prominent image-based rendering methods. It requires the condition of epipolar constraint be satisfied. However, the cylindrical warping involved in the construction of panoramic view makes it highly challenging for this condition to be strictly met. This paper proposes depth-map-based synthesis of virtual environment starting with stereo-panoramic views of a scene. A stereo-panoramic view is first generated by mosaicking a set of images containing both the optical and geometric properties of different parts of the scene. A depth map of the scene contained in the stereo-panoramic view is constructed based on the normalization of matching points and the singular value decomposition of fundamental matrix. Finally, the depth map is textured with one of the panoramic views to obtain a synthesized view. While a good quality depth map can be only generated if and only if epipolar constraint condition for planar rectification is satisfied, it is also the determining factor for the computation of virtual environment from a stereo-panoramic view.
Keywords :
image matching; image texture; stereo image processing; cylindrical warping; depth map texturing; depth-map-based synthesis; epipolar constraint; geometric property; image texture; image-based rendering method; mosaicking; optical property; panoramic view construction; planar rectification; stereo image processing; stereo-panoramic view; virtual environment computation; Stereo vision; image fusion; image registration; image sequences; image texture; stereo image processing;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
3DTV-Conference: The True Vision-Capture, Transmission and Dispaly of 3D Video (3DTV-CON), 2013
Conference_Location :
Aberdeen
ISSN :
2161-2021
Type :
conf
DOI :
10.1109/3DTV.2013.6676659
Filename :
6676659
Link To Document :
بازگشت