al. 2004).
D can be
needs real
ny cases,
resolution
d as the
ate points
rived to
model. In
s limited,
on of two
rocessing
| a typical
formation
y a least
face, and
the finite
ips: (a)
imating
h can be
nal laser
matching
while the
: network
etween
tine.
ordinates
vilities of
out using
Windows
uracy of
essary to
. filtering
technology. And the initial approximation of seven unknown
parameters is used as:
£o ly 7f, 700,9 = @ - K, 200, m, -1.0
In the iteration routine of the least squares solution, the
convergence of rotated angles can be set as le-6 (rad) or le-7
(rad), and the number of iterations may be less than 20.
4.1 Data Description
Two typical data sets were selected in this section. The
scanning data from Hebei and Henan province, China was
received by an ALS60 (Leica Geosystems) airborne laser
scanner at 1000 meters above the ground. The Hebei data set is
an urban scene, including buildings, roads, trees, rivers, etc.
and the Henan data set is a mountainous area, including
terraced field, scarps, isolated trees and a few low buildings.
A
EXE
Figure 4. Two groups of LiDAR raw data sets: (a)~(b) Hebei
data set of urban scene, (c)-(d) Henan data set of
mountainous area.
i
Some other quantitative information of our experiment data
sets can be seen in table 1.
Table 1. Properties of the test data sets
a points number data range overlap points density
1. 00E+04 (m X m) (3) (points/m2)
stripl (hebei) 514. 77 1999 X 1549 17 4.12
strip2(hebei) 477. 12 1997 X 1545 18 4. 30
strip3(hebei) 498. 10 1999 X 1547 17 3.92
strip! (henan) 324. 17 1999 X 2515 15 2.11
strip2(henan) 319. 30 1999 X 2504 14 2.43
strip3(henan) 309. 46 1999 X 2523 13 2.27
4.0 Results
The purpose of this section is to show 2 categories of results
from our proposed method: profiles registering, estimating
results and accuracy assessment of unknown parameters.
Profile results are shown in figure 5. Estimating results of
unknown parameters are listed in table 2 and the corresponding
accuracy assessments are in table 3.
All of the results show that the LS3D proposed in our approach
can achieve the transform parameters between two overlapping
surfaces, and the conjugate points rule plays an important role
in the LS3D procedure.
5. CONCLUSIONS
A general mathematical model for co-registration of two 3D
surfaces is presented. Our proposed method, estimates the
transformation parameters between reference surface and
registration surface, using the generalized Gauss Markoff
model which is a well-known method in geodesy and
photogrammetry. Our mathematical adjustment model is
generic to effective conjugate point rules. In this paper, there
are three definitions of conjugate points presented and
described. Finally, the LND definition shows the highest
precision in the experiments of surface matching or strip
adjustment.
At last, the derived conclusions may be the following ones.
Due to the influence of multiple echo points in laser scanner
data, it is necessary to remove height anomaly points and
irregular geometric shapes, e.g. tree points, using fast filtering
technology or rough classification routines. These points
discussed above can reduce the co-registration accuracy.
Another important issue is that if there are not enough
geometric features or enough overlap between conjugate
surfaces, the iteration will fail. In this case, we can add some
roof points to strengthen geometric constraints between
surfaces.
REFERENCE
Ackermann, F., 1984. Digital image correlation:
performance and potential application in photogrammetry.
Photogrammetric Record, 11(64), pp. 429-439.
Besl, P.J, and McKay, N.D., 1992. A method for
registration of 3D shapes. IEEE Transactions on Pattern
Analysis and Machine Intelligence, 14(2),pp. 239-256.
Crombaghs, M.JE., Brugelmann, R.2000. On the
adjustment of overlapping strips of laser altimeter height data.
International Archives of Photogrammetry and Remote Sensing,
33(B3), pp.224-231.
Chen, Y., and Medioni, G., 1992. Object modelling by
registra-tion of multiple range images. Image and Vision
Computing, 10(3), pp. 145-155.
Gruen, A., 1984. Adaptive least squares correlation —
concept and first results. /ntermediate Research Project Report
to Heleva Associates, Inc., Ohio State University, Columbus,
Ohio, March, pp. 1-13.
Gruen, A., 1985a. Adaptive least squares correlation: a
power-ful image matching technique. South African Journal of
Photogrammetry, Remote Sensing and Cartography, 14(3), pp.
175-187.
Gruen, A., and Akca D., 2005. Least squares 3D surface and
curve matching. /SPRS Journal of Photogrammetry and Remo-
te Sensing 59(3), pp.151-174.
Kornus, W., Ruiz, A. 2003. Strip adjustment of LiDAR data.
International Archives of Photogrammetry and Remote Sensing,
33(3/W), pp.47-50.