Minimal problems in Computer Vision

» Home

» P4P + unknown focal length

» P4P + focal + radial distortion

» uP2P + known vertical direction

» uP3P + focal + radial distortion

» P2P1L problem

» P1P2L problem

» 5-pt relative pose problem

» 6-pt focal length problem

» 6-pt calibrated-uncalibrated

» 4-pt 3-views relative pose problem

» 2-pt panorama stitching (1 focal)

» 3-pt panorama stitching (2 focals)

» 3-pt panorama stitching (1 focal+ radial distortion)

» 6-pt generalized camera problem

» 6-pt calibrated radial distortion

» 8-pt uncalibrated radial distortion

» 9-pt different distortion problem

» P2Π registration problem

» 3-view triangulation

» 9-pt catadioptric

» Automatic generator

» Grobener basis solver

4-pt 3-views relative pose problem

Problem:

Estimating relative pose of three fully-calibrated cameras from four image point correspondences.

Publications:

D. Nister and F. Schaffalitzky. Four points in two or three calibrated views: Theory and practice. International Journal of Computer Vision, Volume 67, Number 2, April, 2006. [pdf | http]
L. Quan, B. Triggs, and B. Mourrain. Some results on minimal euclidean reconstruction from four points. J. Math. Imaging Vis., 24(3)::341–348, 2006. [pdf | http]
H. Li. Multi-View Structure Computation without Explicitly Estimating Motion, in IEEE CVPR 2010. [pdf | http | code]

Code:

Evaluation:

Tomas Pajdla, Zuzana Kukelova, Martin Bujnak. Maintained by Zuzana Kukelova, 28 January 2011