Determining relative geometry of cameras from normal flows

Authors:
Ding Yuan;Ronald Chung
Affiliations:
Department of Mechanical & Automation Engineering, The Chinese University of Hong Kong, Shatin, Hong Kong, China;Department of Mechanical & Automation Engineering, The Chinese University of Hong Kong, Shatin, Hong Kong, China
Venue:
ACCV'07 Proceedings of the 8th Asian conference on Computer vision - Volume Part II
Year:
2007

Citing 7
Cited 0

A theory of self-calibration of a moving camera

International Journal of Computer Vision
Qualitative egomotion

International Journal of Computer Vision - Special issue: image understanding research at the University of Maryland
Geometric Camera Calibration Using Circular Control Points

IEEE Transactions on Pattern Analysis and Machine Intelligence
Real-Time Epipolar Geometry Estimation of Binocular Stereo Heads

IEEE Transactions on Pattern Analysis and Machine Intelligence
Self-Calibration of a Stereo Rig in a Planar Scene by Data Combination

ICPR '00 Proceedings of the International Conference on Pattern Recognition - Volume 1
Direct estimation of the stereo geometry from monocular normal flows

ISVC'06 Proceedings of the Second international conference on Advances in Visual Computing - Volume Part I
Stereo geometry from 3D ego-motion streams

IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics

Quantified Score

Hi-index	0.00

Visualization

Abstract

Determining the relative geometry of cameras is important in active binocular head or multi-camera system. Most of the existing works rely upon the establishment of either motion correspondences or binocular correspondences. This paper presents a first solution method that requires no recovery of full optical flow in either camera, nor overlap in the cameras' visual fields and in turn the presence of binocular correspondences. The method is based upon observations that are directly available in the respective image stream - the monocular normal flow. Experimental results on synthetic data and real image data are shown to illustrate the potential of the method.