Introduction
PBA is a multi-view
feature matching software that can be used to improve camera calibration in
combination with our multi-view stereo software (PMVS)
and a standard bundle adjustment software such as
sba by Manolis Lourakis
and Antonis Argyros. More concretely, given a set of images observing a
rigid object or a scene (possibly containing obstacles such as pedestrians
in front of buildings) and the associated camera parameters that may contain
some errors (see the documentation for how
much errors are allowed),
1. you first use PMVS with the input images and the initial camera
parameters to reconstruct oriented points;
2. second, use PBA software to match features in multiple images by using
reconstructed oriented points;
3. third, use sba (or any standard bundle adjustment software) to refine
camera parameters based on the identified feature correspondences.
In practice, you can iterate these three steps to improve camera
calibration. Please see our documentation page
for the details.
Our PBA software can
be downloaded from here (binary programs for Windows). Besides PBA, you need
two more software as mentioned above. Our PMVS software is available
here
(we provide PMVS to only people in academics for research purposes, and ask
you to agree with the terms and conditions). Sparse bundle adjustment
software (sba) can be downloaded from
here.
The algorithm implemented in the software
(pba) is described
in our
CVPR 2008 paper.
Contacts
For
comments or bug reports, you can write to furukawa -at- cs.washington.edu.
Please use "bundle adjustment" as a
subject of an email. However, we do not provide any support or
help regarding the software beyond the documentation, as we are
getting many emails and will not have time to answer each one
of them.