The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B5. Beijing 2008 
618 
on the local incidence angle. Angles plotted in red may indicate 
scan points of less accuracy. It is clearly seen that for large in 
cidence angles, the point cloud density decreases. Note that at 
larger incident angles (* > 60°), the intensity measurements for 
the wooden plate and the test charts are distorted, relative to the 
stand-point A intensity measurements (Fig.2(a)). This effect is 
due to bigger footprints for high incidence angles. 
4.3 Segmentation 
Fig.4 depicts the segmentation result for both stand-points. The 
room is segmented into eight main segments. As depicted in 
Fig.4, at larger incident angles (i > 60°), the wooden plate seg 
ments and the lower wall test chart segment are distorted and big 
ger, when comparing to the stand-point A segmentation (Fig.2(a)). 
This example clearly demonstrates that changing local point qual 
ity has immediate effects on post-processing, like in this case seg 
mentation. 
4.4 Individual point residual of a planar fitting 
For each segment, a plane is fitted according to the method de 
scribed in Sec.2.6. Fig.5 shows the residual e for each point of 
each segment of the point cloud. The points colored in magenta 
represent residuals higher than 2 cm. Fig.5 shows the individual 
point residual for the stand point A. The estimated planes for 
both walls produce very low residuals (< 1 cm). The points on 
the floor segment that are situated in high incidence angle areas 
produce high residuals. By moving the scanner from the stand 
point A to the stand-point B, high incidence angles at the bottom 
left comer of segment 1 are avoided. In white, the RMSE per seg 
ment is plotted. The differences in RMSE between stand-point A 
and B can partially be explained by comparing the different inci 
dence angle pattem.for each segment. Note that in the stand-point 
B results, a stripe can be observed in segment 1, corresponding 
to the 0 = 360° transition of the horizontal scan angle. A 
possible explanation for this effect can be found in saturation: in 
Fig.2(b), a saturation effect is observed at the near perpendicular 
surfaces to the laser beam. The points measured shortly after the 
saturation are all affected by a higher residual. This effect can 
be explained by an overload of the intensity sensor of the laser 
beam. 
4.5 Patch point density 
In Fig.6, the number of points per patch of 20x20 cm is shown. 
In general, the scanner position at the stand-point B results in 
a higher point density. For both positions, the point density de 
creases rapidly with range and with increasing incidence angle. 
This holds especially towards the far sides of the segments. 
4.6 Planar patch quality 
Each segment is subdivided into small patches of 20 x 20 cm. 
From the points in the patch, local planar parameters are deter 
mined as described in section 2.7. The quality of the local patch 
is evaluated by one number: the standard deviation in the direc 
tion perpendicular to the local patch of the center of gravity of 
the points belonging to the patch. This standard deviation a m 
is determined according to Eq.8. Clearly this standard deviation 
reflects both the individual point quality, compare Fig.5, and the 
local point density, compare Fig. 6. In Fig.6 the mean of the 
patch variances for each of the four largest segments is plotted 
in white. On average, the position in the comer (stand-pointS) 
results in patches of better quality. The average patch variance 
for all patches together equals 0.0023 m for the stand-point A 
and 0.0017 m for the stand-point B. This shows that by simply 
moving the scanner by two meters, the quality of the point could 
can be improved by 25 %. 
5 CONCLUSIONS AND FUTURE WORK 
It is well-known that the position of the scanner affects the quality 
of individual scan points. In this paper, we were able to actually 
quantify this effect by introducing a notion of point cloud quality, 
that incorporates both the point density and the individual point 
quality. It is shown that by moving a scanner by an ample 2 me 
ters, the point cloud quality can be improved by 25 %. 
In a next step the optimal scanner position could be determined 
by using error models from the major error components, like the 
scanning geometry, the material properties, the scanner mecha 
nism and the environmental conditions. For a complex scene, first 
a small resolution sketch scan could be used to determine the op 
timal measurement setup. To proceed in this direction a thorough 
knowledge of all the error components is preferable, meanwhile 
use can be made of well described parameters like the incidence 
or the point density. 
REFERENCES 
Böhler, W., Bordas Vicent, M. and Marbs, A., 2003. Investigating Laser 
Scanner Accuracy. In: CIPA 2003 XVIII International Symposium, Vol. 
XXXIV(5/C15), Antalya,Turkey, pp. 696-701. 
Bucksch, A., Lindenbergh, R. and van Ree, J., 2007. Error budget of 
Terrestrial Laserscanning: Influence of the intensity remission on the scan 
quality. In: Proceedings GeoSiberia, Novosibirsk, Russia. 
Cheok, G. S., Leigh, S., and Rukhin, A., 2002. Technical report: Calibra 
tion Experiments of a Laser Scanner. National Institute of Standards and 
Technology. 
Clark, J. and Robson, S., 2004. Accuracy of Measurements made with 
a CYRAX 2500 Laser Scanner Against Surfaces of Known Colour. 
In: XXth ISPRS Congress, Geo-Imagery Bridging Continents, Vol. 
XXXV(B5), Istanbul, Turkey, pp. 1031-1036. 
FARO, 2007. Laser Scanner LS 880 Techsheet. Accessed September 
2007, http://www.faro.com/. 
Gorte, B., 2007. Planar Feature Extraction in Terrestrial Laser Scans 
using Gradient Based Range Image Segmentation. In: Proceedings of 
the ISPRS Workshop, Laser Scanning 2007 and SilviLaser 2007, Vol. 
XXXVI(3AV52), Espoo, Finland, pp. 173-177. 
Kiemen, T., Koska, B. and Pospßil, J., 2006. Verification of Laser Scan 
ning Systems Quality. In: Proceedings XXIII FIG Congress, Munich, 
Germany. 
Lay, D. C., 2002. Linear Algebra and Its Applications (3rd edition). 
Addison-Wesley. 
Li, W. X. and Mitchell, L. D., 1995. Laser scanning system testing-Errors 
and improvements. Measurement 16(2), pp. 91-101. 
Lichti, D., 2007. Error modelling, calibration and analysis of an AM-CW 
terrestrial laser scanner system. Vol. 61-5, pp. 307-324. 
Lichti, D. and Gordon, S., 2004. Error propagation in directly georefer- 
enced Terrestrial Laser Scanner point clouds for cultural heritage record 
ing. In: Proceedings FIG WSA Modelling and Visualization, Athens, 
Greece. 
Lindenbergh, R., Pfeifer, N. and Rabbani, T., 2005. Accuracy Analy 
sis of the Leica HDS3000 and feasibility of tunnel deformation monitor 
ing. In: Proceedings of the ISPRS Workshop, Laser scanning 2005, Vol. 
XXXVI(3/W3), Enschede, The Netherlands, pp. 24-29. 
Pfeifer, N., Dominger, R, Haring, A. and Fan, H., 2007. Investigating 
terrestrial laser scanning intensity data: quality and functional relations. 
In: Proceedings International Conference on Optical 3-D Measurement 
Techniques VIII, Zurich, Switzerland. 
Soudarissanane, S., Van Ree, J., Bucksch, A. and Lindenbergh, R., 2007. 
Error budget of terrestrial laser scanning: influence of the incidence angle 
on the scan quality. In: Proceedings 3D-NordOst 2007, Berlin, Germany. 
Teunissen, R, 1991. An integrity and quality control procedure for use 
in multi sensor integration. In: Proceedings of ION GPS-90, Colorado 
Springs,US A, pp. 513-522. 
Zhuang, H. and Roth, Z. S., 1995. Modeling gimbal axis misalignments 
and mirror center offset in a single-beam laser tracking measurement sys 
tem. The International Journal of Robotics Research 14(3), pp. 211-224.