art B3. Istanbul 2004
ETWEEN
DETIC SYSTEM
rs of the buildings is
of the image utilizing
nputer aided design).
ations between the
volved, one proceeds
le generated in the
atical model performs
coordinates (xp, yp) of
local tri-dimensional,
1), utilizing inverse
ransformation may be
V: HITA, (1997).
- Xo)( yf — yo) (8)
0f-y)) (0
ystem;
'stem;
togrammetric system;
rameters;
21S;
eters;
t in the image system;
station;
| geodetic system;
().R(m)).
srential was utilized as
applications with the
tions, the value of the
coordinate Z, is obtained using an iterative process
supported by a digital terrain model (DTM). In this
application where the objective is the mapping of
buildings with the integration of laser scanner data, a new
procedure to obtain this coordinate was implemented. The
points originated from the laser scanner survey that define
the borders of buildings are projected on the image space
with the collinear direct equations, as presented in figure
02 and 03. A group of (7?) points with coordinates on the
hybrid system
geodetic) (x
(photogrammetric and
pis Y pi» 7 =1,n is generated. The
photogrammetric point of a corner of a building is
observed on the referential of the image. After simple
mathematical transformations, the coordinates on the
photogrammetric referential are determined. To determine
the local geodetic coordinates [X,Y,] with the
application of inverse collinear equations, the value of the
coordinate Z, must be determined. Utilizing the nearest
neighbor interpolation technique, the value Z; of the
point to be rectified is determined in the files of
coordinates of transformed laser scanner points
(pis Vir Ly, zl.
The proposed methodology was implemented in the 3D
MONOPLOTTER computational program (See Fig. 04)
developed at the Graduate Program in Geodetic Sciences,
UFPR.
Figure 4: Main page of 3D Monoplotter program.
7. RESULTS OBTAINED IN THE
MONORESTITUTION OF BUILDINGS
7.1 Exterior Orientation
Employing collinear equations in direct form and MMQ
adjustment (Least Squares), the parameters of exterior
orientation of the image were determined. To conduct the
adjustment, eight control points distributed all over the image
were observed. Coordinates of these points in the
photogrammetric system were obtained from observation
conducted in the digital image with the CAD MicroStation PC
System and the coordinates in the local geodetic system
obtained from the observation of photogrammetric models in
the analytical photogrammetric ZEISS PLANICOMP C-100,
as described in item 5.0 of this paper. The main results
obtained in the adjustment are:
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
Kappa (x) 7 1.9651533575 64, — 0.0005451726
Phi (9) 7 0.0064097357 o = 0.0027718695
Omega (9) = -0.0158918124 Gr) =0.0030046261
Xo(m) = 108.6212222420 © xo) = 2.1647013270
Yo(m) = -76.2077762045 © y,) = 2.3494253375
Zo(m) = 1652.6215413585 © zo) = 0.4449289549
Standard deviation of the residues in the photogrammetric
coordinates: (O «, — 0.003 mm e O , = 0.003 mm)
Standard deviation of the residues in the local geodesical
coordinates: (O x) = 0221 m; O wv = 09.191m; Oz, =
0.063 m)
7.2 Tridimensional Monoplotting
Utilizing the 3D Monoplotter Program, the digital vector file
was rectified according to the methodology presented in item
6.0. The tridimensional coordinates of some corners of the
buildings (X,Y and Z) were obtained in the vector rectified
file. The planimetrical (X, Y) coordinates were determined
with the application of inverse collinear equations and the
altimetric coordinates of the interpolation performed with the
laser scanner data. The values obtained from the rectified
vector file were compared with the coordinates determined
with the reading of photogrammetric models in the ZEISS
PLANICOMP C-100 System, as described in item 5.0 of this
paper. To verify the accuracy of the 3D monorestitution
proposed in this research, it was considered that the
tridimensional coordinates of the corners of the buildings
obtained in the PLANICOMP C-100 are correct, exempt from
error of observation and others connected with the
photogrammetric process utilized.
Table 2 shows the accuracy results obtained in the 3D
monorestitution in the region of large buildings. The acronyms
DX, DY and DZ are discrepancies in meters in the three tri-
dimensional coordinates of the point and Dpla is the
planimetric discrepancy existing.
Figure 5 shows the spatial distribution of the planimetric
discrepancies presented in Table 2. One can verify that the
planimetric discrepancies are well distributed around the origin
0.0 and 90% of the points tested present planimetric accuracy
of up to 0.5 of the meter. The results obtained for the 3D
monorestitution of big buildings are equivalent to conventional
stereophotogrammetric restitution in the scale of 1/2000.
In table 3 and figure 6, the accuracy obtained with the 3D
monorestitution in the region of small buildings is presented.
In this case, the accuracy obtained in the determination of the
tri-dimensional coordinates is inferior if compared with the
former case. One can verify that 90% of the points tested
present planimetric precision of up to 0.70 of the meter. This
smaller precision is related to two main deficiencies. The first
is linked to higher degree of difficulty to define the points that
represent the borders of the buildings in the group of points
proceeding from the laser scanning, as mentioned in item 4.0,
and the second, the difficulty to identify the corners of
edifications in the digital image, due to the resolution of the
image.
One can verify in figure 5 a small tendency of systematic
spacing in the determined planimetric discrepancies, of
