The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B5. Beijing 2008 
the local patch parameters. In this paper, the point cloud is repre 
sented as a net view, therefore, the only planar parameter quality 
to be estimated is the height z of the point with respect to the 
planar fitting. Assume the following linear model: Z = A ■ P, 
where A = [x», t/i, l]t=i,..., n , Z = [zi]i=ri,..., n and P represents 
the planar patch parameters. The variance-covariance matrix of 
the planar patch parameters is given in Eq.7 
Q, = A T Ql'A 
(7) 
In determining the local patch parameters, a higher number of 
local patch points will result in more accurate local patch param 
eters. In general, using the redundancy of the observations allows 
to derive adjusted points on the adjusted local planar patch with a 
precision far below the nominal point precision of an individual 
laser point. For each set of points X, the propagated variance a rn 
at the center of gravity M is considered as shown in Eq.8. 
O’rn — A-mQf) 
(8) 
where A m = [M x ,M y , 1]. 
3 EXPERIMENT SET-UP 
The laser scanner measurements optimization is investigated us 
ing the experiment set-up as shown in Figure 1. The laser scanner 
LS880 HE80 from FARO (FARO, 2007) is used. The laser beam 
of this laser scanner is deflected at 90° on a rotating mirror which 
determines the vertical field of view of 320° since the scanner 
cannot scan under itself. The head of the scanner rotates around 
its vertical axis to allow the horizontal field of view of 360°. A 
full resolution scan has typically around 130 million of points. 
The experiments are performed in a closed area with short ranges, 
therefore the temperature and humidity influences are neglected. 
The scans considered here contain about 26 millions of points. 
The room scanned for this experiment consists of two planar 
walls and one cylindric wall. As the focus of this paper is into 
the planar features quality, the cylindric wall is excluded from 
the analysis. As depicted in Fig. 1(a), the laser scanner provides a 
panoramic view of the area by measuring the reflection of a phase 
modulated laser beam. The laser scanner cannot scan shiny ma 
terials such as metal or mirror like materials, and low reflectance 
materials are measured with lower accuracies (Bucksch et al., 
2007, Kremen et al., 2006, Clark and Robson, 2004). The ceil 
ing of the room of experiment contains very shiny materials and 
is composed of several small segments. Therefore, the ceiling 
is not part of this study. The floor is covered with light colored 
linoleum. The walls are painted in white and are very smooth 
surfaces. 
Four test plates that were used in previous studies are added on 
the two planar walls. Two reference charts (ESSER TE 106 and 
TE 109) for color and grey scale were previously used in a remis 
sion experiment (Bucksch et al., 2007). A white coated plywood 
and a medium-density fibre board were used before in a scan an 
gle experiment (Soudarissanane et al., 2007). Fig. 1(b) represents 
a 3D model of the room of experiment. 
The laser scanner scans the room from two different stand-points. 
The stand-point A is approximately situated in the middle of the 
room. The stand-point B is situated in the comer formed by the 
two planar walls. 
4 RESULTS AND DISCUSSION 
In this section, the influence of the location of the laser scanner on 
the point cloud quality is presented, based on two stand-points. 
(a) 
(c) 
Figure 1: (a) Panoramic intensity image obtained with the FARO 
LS880 laser scanner, (b) 3D view model of the experiment set 
up, (c) Net model of the room of experiment, A and B are two 
stand points of the laser scanner. 
4.1 Intensity measurements 
In addition to the cartesian coordinates, for each point in the point 
cloud an intensity value ranging from 0 to 1 is given. This in 
tensity value represents the amount of reflected light intensity as 
regard to the emitted light. This value is provided by the manu 
facturer of the laser scanner. According to the providers, the laser 
scanner measure the received intensity value, which depends on 
the surface roughness, but also on the scattering behavior of the 
surface based on its reflectivity properties. Note that this prod 
uct is not calibrated. Fig.2 depicts the point cloud colored with 
the measured intensity value for both stand-points. As the laser 
scanner cannot scan under itself, an empty spot is observed at the 
position of the laser scanner. 
The intensity values on the walls at the position A, shown in 
Fig.2(a) are homogeneous for each scanned surface. The dis 
tances to each surface are large enough to obtain homogeneous 
intensity values. The walls are painted in white, which has high 
reflectance properties. The returned signals are stronger for the 
white walls than for the light-colored floor reflections. The wooden 
plate hang on the upper left wall as depicted in Fig.2(a). It has 
lower reflectance properties than the white walls or the white 
plate, therefore has lower measured intensity values. The two 
reference charts are having a black-colored frame with a very low 
reflectance property. 
Fig.2(b) shows the intensity measurements from position B, where 
the laser scanner was placed nearer to the comer formed by the 
two planar walls. A saturation effect is observed for signals ob 
tained with near perpendicular scanning direction. The white 
walls and the light-colored floor have a similar order of inten 
sity values. At the near perpendicular directions, the saturation is 
characterized by very high intensity values. The spatial intensity 
distribution is clearly affected by the position of the scanner. 
4.2 Incidence angles 
Fig.3 depicts the incidence angle of the laser beam for each stand 
points. Clearly the position of the laser scanner has an influence 
617