Summary: 3D STRUCTURE FROM VIDEO STREAMS WITH PARTIALLY OVERLAPPING IMAGES
Rui F. C. Guerreiro Pedro M. Q. Aguiar
Institute for Systems and Robotics, IST, Lisboa, Portugal
The majority of methods available to recover 3D structure from
video assume that a set of feature points are tracked across a large
number of frames. This is not always possible in real videos be-
cause the images overlap only partially, due to the occlusion and
the limited field of view. This paper describes a new method to re-
cover 3D structure from videos with partially overlapping views.
The well known factorization method  recovers 3D rigid
structure by factoring an observation matrix that collects trajecto-
ries of feature points. We extend this method to the more challeng-
ing scenario of observing incomplete trajectories. This way, we ac-
commodate not only the features that disappear, but also features
that, although not visible in the first image, become available later.
Under this scenario, the observation matrix has missing entries.
We develop three new algorithms to factor out matrices with miss-
ing data. Experiments with synthetic data and real video images