Chapman, David
Range-error (mm)
O ND Wh OO NN Oo do
500 1000 1500 2000 2500 3000 3500
Stand-off Distance (mm)
SE
Figure 1. The Long Range Biris sensor Table 1. Range error for the Lu range Biris as a function of stand-off
istance
À calibration structure, covering the views of the Biris scans (partially shown
in figure 2), is used to calibrate the 3-D images produced by the Biris and the
scanning pan-tilt unit (see El-Hakim et al, 1997, for details). The spherical
targets on the structure are surveyed to 0.08 mm accuracy. The estimated
accuracy numbers using the targets on the calibration structure as reference
are:
RMS(X): 2.27 mm
RMS(Y): 2.44 mm
RMS(2): 1.87 mm
The results are close to the theoretical expectations computed from sensor
resolution and the mathematical model (Table 1). RMS values are obtained
from differences between known and computed coordinates.
"ms Thus the system was felt to be capable of delivering range data of sufficient
| coverage and precision to enable the construction of a crude CAD model
suitable for clash determination. This system could be easily integrated within
an existing panoramic photogrammetric measurement system that would
provide high quality image overlays and the capability to selectively upgrade
OM local measurements. This photogrammetric module would also enable the
Figure 2: The calibration structure. ^ measurement of small bore piping and cabling that was too small to be
resolved in the range-image data.
3 SURVEY OF THE TEST ENVIRONMENT.
Since it was not possible to fully integrate the photogrammetric and range-imaging systems in the budget and timescale
of this trial they we deployed separately and brought into a common coordinate framework through the measurement of
a number of targeted control points. Figure 3 shows low resolution panoramas generated by the two systems. The
partial Biris panorama is generated from four overlapping cylindrical strips each strip comprising 256 x 1024 pixels. A
control survey was based upon 11 panoramic photogrammetric stations acquired and localised using the Hazmap
remote measurement system (Chapman & Deacon, 1997) with Stereo panoramas being acquired at 4 stations for
visualisation purposes. Range images were acquired at 10 locations thus 40 strips of range image data were available
for subsequent analysis.
124 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000.
4 DE
As can
geomet
segmer
were d
extracti
e Re
e Ra
e Ra
Of thes
The me
At pre:
being |
maxim
41 F
The ge
Since
requiri
are act
4.2 I
Cylind
param
least-s
initial