2004
nts of
r and
in a
'nsors
aser).
ferent
eived
ignals
ystem
is to
(GPS
SS; ds
data
1: the
ich is
| time
To is
sm or
> but
1, the
| and
ensor
) sets
ation
n the
two
on to
s the
point
neras
P the
n the
1 the
)BIL
on is
ation
'nced
ages
f the
owed
1etric
was
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004
considered too expensive and the actual system is not stereo.
For each pair, each image is visualized in a view and the
operator should identify the point to be measured in both
images. Vector data can be superimposed on top of the images.
The system is based in MicroStation 95 and the customization
has been developed using MDL language. The system
functionality for image and vector visualization, data
digitalization and editing or data storing takes advantage of the
MicroStation basic tools. As the same tools are used in data
compilation, editing and storing for other topographic databases
created at the Institut Cartografic de Catalunya, data integration
can be achieved without any data transformation.
The application allows the reproduction of the path covered by
the GEOMOBIL, visualizing the sequence of images collected
by each camera. The path information is obtained from a file
with the position (geographic coordinates in a given reference
system) and attitude of the projection center for every photo.
Another file provides information about the parameters for the
cameras (focal length, principal point coordinates, radial
distortion and internal orientation).
Each image sequence is visualized in a MicroStation view. The
view is configured according to the camera (coordinates and
angles of the projection center, coordinates of the principal
point, and focal length), and the image is placed in the
perpendicular plane to the camera axis at a selected distance of
the projection center. This method allows the visualization of
georeferenced data on top of the image.
Two views of MicroStation can be used to show the path, the
projection centers and the orientation of the visualized images.
Vector data and raster images can be also displayed in these
views. As no more than eight views can be used in
MicroStation, a maximum of six sequences can be visualized
during a session.
ue Ne ttc 95
ET n
i 1V-LTA4 LC-SDUCU-9 TP Cebtes «
Figure 4: Screen shot of the GEOMOBIL data
extraction software.
The tools for image management allow to advance and to go
backwards one image in one or more views, to select one image
in one view, to synchronize the images for all the views, and to
go to the nearest image to a georeferenced point. The
visualization tools, as zoom in, zoom out, center the view to a
given point, or fitting the image in the view, have been
specially developed for this application, because MicroStation
standard visualization tools do not preserve the camera
configuration.
The point coordinates are calculated resolving the collinear
equations {rom the position identified in two images. The
images can be amplified to facilitate the point identification,
and the epipolar line can be visualized, after giving the point in
the first image, to identify the same point in the second image.
The collinear equations solution is computed in geocentric
coordinates to increase the accuracy, and then it is converted to
the work reference system.
The coordinates of the calculated point are sent to MicroStation
to be processed by the active command as a standard input.
Microstation commands and other specific tools can be used to
capture new data or to modify existing information. A set of
tools has been created to gather elements tied in the roads. They
allow to capture the axis of the route, to attribute it with some
characteristics (number of rails, tunnel, street...) and to capture
attached elements to the road as kilometric points, traffic signs,
gas stations or bus stops.
3. GEOMOBIL IMAGE SUBSYSTEM CALIBRATION
3.1 Camera parameters calibration
Calibration of the optical parameters was carried out at the [CC
facilities.
The floor of the ICC exposition room has a regular pattern (as
can be seen in image 5). Using classical surveying techniques, a
local reference frame was set in the exposition room. The
coordinates of six points on the ground in the local reference
frame were measured. The position of each camera was
computed on up to six different sites around the target area on a
balcony situated at 8.40 meters above the target area. From each
of these sites both cameras were operated to image the target
area.
Figure 5: setting of the camera calibration site.
Up to 471 points were identified in the images of the target
area, with in a total amount of 4165 photogrammetric
observations (average of 347 photogrammetric observations per
image). Six of the 471 points are the six measured points
mentioned above. These six points become six full control
points. The other 465 are vertical control points at height zero
(in the local reference frame). Moreover, due to the regular
pattern and distribution of points, a distance restriction between
the adjacent points of the 471 point network was imposed.