jan
sis
ch.
ter
ity
07,
lli-
ific
44,
ing
age
iter
age
yer
LOS,
ine
für
age
85—
Vittorio Casella
ESTIMATION OF PLANIMETRIC ACCURACY OF LASER SCANNING DATA. PROPOSAL OF
A METHOD EXPLOITING RAMPS
Vittorio CASELLA, Anna SPALLA
University of Pavia, Italy
Department of Building and Territorial Engineering
casella Q unipv.it
spalla ?unipv.it
Working Group III/1
KEY WORDS: Accuracy, Algorithms, Data processing, DTM/DEM/DSM, Laser/Lidar.
ABSTRACT
A method is described for the estimation of planimetric precision of laser scanning data. Due to the nearly random dis-
tribution of the points hit by the laser, it is difficult to use the very standard technique of measuring the same points with
the instrument under investigation and with another one, whose performances are well known and better than those
expected from the former. But when the terrain is not flat, it could be said that the planimetric errors affect altimetric
precision of the points measured by the laser scanner; we will show exactly what this means. We will also show that
this behaviour can be exploited to estimate planimetric precision, provided that the shape, position and attitude of some
ramps is known. The paper first of all focuses on the method's description, then it shows some initial results obtained
from Pavia's test area.
1 INTRODUCTION
Laser scanning is rather a new technique which is already widely used. Nevertheless, it hasn't yet been completely in-
vestigated for some aspects, such as the accuracy of the sparse (X,Y,Z) data. Altimetric accuracy can be easily esti-
mated on flat areas such as car parks, sport fields, squares and road crossings. Planimetric accuracy, instead, cannot be
easily assessed by usual methods, based on the comparison of measurements given by different techniques. Indeed laser
sensor scans the world below blindly and the points measured by it do not necessarily coincide with features usually
exploited by surveyors such as building corners or edges. Therefore it is almost impossible to measure the same points
with the laser and with another instrument, such as GPS.
One solution is represented by interpolation. Laser data could be interpolated to obtain grids and these could be ana-
lysed to find building edges and to determine their position. These positions could finally be checked by means of direct
measurements. But unfortunately the gridding process and the following edge extraction introduce into the data some
errors which are not easily distinguishable from the errors originally affecting the laser data. Therefore, even if gridding
and grid analysis are well known and powerful methods, we tried to develop another way of assessing planimetric pre-
cision.
The method suggested by us does not require gridded data and it is based, instead, on sparse points. It exploits planar
man-made surfaces such as court ramps and river banks. Let's consider first of all a simplified ramp and let's suppose
that a laser pulse has struck a point whose true coordinates are (X,.Y,,Z,). Due to the measurements errors, the sensor
will return, for the hit point, the position (X,Y,Z).
The relationship between the real, unknown, position and the measured one involves random errors (and systematic
ones, if any)
(X,Y,Z )2 (X, Ys Z9 * (ex.e. €) (1)
Once we have measured the ramp with a high-precision method, such as GPS or a mixture of it with other land surve y-
ing methods, it is possible to estimate the parameters a, b and c of the plane with a least squares approach; then, it is
possible to compare the measured height, Z, and the height of the ramp at the same location, given by aX + bY+c.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 157