Graphics Reference
In-Depth Information
been estimated as one of the state-of-the-art MVS algorithm [9] as automatically ge-
nerating an accurate, dense and robust result. Poisson Surface Reconstruction pro-
posed in [10] will be used as a post-processing step to convert the set of oriented
points produced by PMVS into a triangulated mesh model. Fig.4 shows an effect pic-
ture of a screen wall in the Qiao's Grand Courtyard reconstructed by this method.
Fig. 4.
Effect of reconstructed models in the system
Multi-view stereo (MVS) matching and reconstruction is a key ingredient in the
automated acquisition of geometric object and scene models from multiple photo-
graphs, a process known as image-based modeling or 3D photography. Here we util-
ize a classic algorithm for multi-view stereopsis that outputs a dense set of small
rectangular patches covering the surfaces visible in the images.
Stereopsis is implemented as a match, expand, and filter procedure, starting from a
sparse set of matched key points, and repeatedly expanding these before using visi-
bility constraints to filter away false matches. The keys to the performance of
the proposed algorithm are effective techniques for enforcing local photometric
consistency and global visibility constraints.
A patch
p
is essentially a local tangent plane approximation of a surface. Its center
and normal respectively denoted as
c
(
p
) and
n
(
p
)
. A reference image
R
(
p
)
is the pic-
tures used in the matching step. A patch is a 3D rectangle, which is oriented so that
one of its edges is parallel to the x-axis of the reference camera (the camera associated
with
R
(
p
)
). The extent of the rectangle is chosen so that the smallest axis-aligned
square in
R
(
p
)
containing its image projection is of size
ʼ
×
ʼ
pixels in size. Let
V
(
p
)
denote a set of images in which p is visible, I can refer to any images,
V
*(
p
)
is the
image set ignoring images with bad photometric discrepancy scores by simply adding
a threshold.
cpOI
()()
(3)
Vp
() {|()
=
Inp
⇅
>
cos()}
˄
cpOI
()()
Vp
*(
)
=
{
IIVphpIRp
ʱ
|
∈
(
),
(
,
,
(
))
≤
}
(4)
The photometric discrepancy function
g
(
p
) for
p
on
V
(
p
)
and
V*
(
p
)
is defined as
Search WWH ::
Custom Search