Information Technology Reference
In-Depth Information
3.5
4
Roll
Pitch
Yaw
Epipoles
Virtual Epipoles
3.5
3
3
2.5
2.5
2
2
1.5
1.5
1
1
0.5
0.5
0
0
0
0.4
0.8
1.2
1.6
2.0
0
0.4
0.8
1.2
1.6
2.0
Noise std. dev. [pixels]
Noise std. dev. [pixels]
(a)
(b)
0.35
Sylvester equation
Screw axis
0.3
0.25
0.2
0.15
0.1
0.05
0
0
0.4
0.8
1.2
1.6
2.0
Noise std. dev. [pixels]
(c)
|
θ
−
θ
|
Fig. 1.8
Simulations: (a) mirror angle estimation error
for the algorithms in Proposi-
tions 1.5 and 1.6; (b)
R
c
w
estimation error for the algorithm in Proposition 1.7; and (c) error
ε(
t
w
) for the algorithms in Proposition 1.8
scale-invariant feature transform (SIFT) [12] and, in a second phase, assign the
points to each mirror which has been uniquely identified by a colored landmark.
The rigid motion (
R
c
w
,
t
c
w
) between the camera and world frame is
R
c
w
=
R
z
(40 deg)
R
y
(
t
c
w
=[0
2]
T
−
20 deg)
R
x
(30 deg)
,
.
6
−
2
−
0
.
m
.
Four well-known fundamental matrix estimators [1],
normalized 8-point algorithm
,
iterative Sampson method
,
robust Torr's M-estimator
and
LMedS
, have been com-
pared in the simulation experiments. The M-estimator revealed the best compromise
between estimation accuracy, computation load and robustness to uncorrect corre-
spondence matching due to outliers.
Search WWH ::
Custom Search