Graphics Reference
In-Depth Information
Ta b l e 5 . 6
Estimated pose
and ground truth (GT) for the
rubber example
Parameter
Pose 1
GT 1
Pose 2
GT 2
Roll (
◦
)
13
.
3
13
.
5
16
.
7
16
.
3
Pitch (
◦
)
−
18
.
2
−
18
.
9
−
18
.
6
−
19
.
7
Ya w (
◦
)
59
.
4
58
.
6
59
.
2
58
.
5
t
x
(mm)
−
3
.
6
−
3
.
2
−
2
.
8
−
2
.
5
t
y
(mm)
2
.
3
2
.
3
1
.
3
1
.
7
t
z
(mm)
451
.
5
454
.
3
457
.
5
453
.
9
5.6.6 Discussion
The three-dimensional pose estimation approach described in this section is based
on photometric, polarimetric, edge, and defocus cues. A correspondingly defined er-
ror function is minimised by comparing the observed data to their rendered counter-
parts, where an accurate rendering of intensity and polarisation images is performed
based on the material-specific reflectance functions determined with a goniometer.
If a certain cue cannot be reliably measured or does not yield useful information, it
can be neglected in the optimisation procedure.
A pose estimation accuracy comparable to the one obtained in the simple rubber
example is achieved for more difficult objects in the context of industrial quality
inspection, as described in Sect.
6.1
. It turns out that this accuracy is comparable to
or higher than that of the monocular template matching approach of von Bank et al.
(
2003
) (cf. Sect.
2.1
), which exclusively relies on edge information and estimates
only five degrees of freedom. The depth from defocus method has proven to be a
useful instrument for the estimation of object depth in close range at an accuracy of
about 1 %.
Search WWH ::
Custom Search