' - '¿¡.-y 7? _ Lrs
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008
Figure 6: Pedestrian crossing
From each pedestrian crossing, a resolution measurement is
extracted, taking the mean of several values obtained over
different parts of the crossing, in order to minimise the effects
of noise. Furthermore, frames with a high number of crossings,
placed horizontally and vertically, have been chosen in order to
compare the results that we obtain with the theoretical
estimations that we already have.
The Figure 7 and 8 show the value of resolution obtained from
the application of the software in different pedestrian crossings,
based on the distance to the centre of the shot, compared to the
theoretical values of resolution in flight and cross-flight
direction, respectively.
Figure 7: Resolution cross-flight direction
Figure 8: Resolution flight direction
7. CONCLUSIONS
DMC digital images prove to have a big improvement in terms
of image correlation reaching a matching accuracy of 1/10 th of a
pixel. In aerotriangulation check points showed that an
accuracy of 1/2 pixel in the horizontal components and 2/3 of a
pixel in the vertical component can be routinely archived in
production flights. In this study, however, the high accuracy in
the correlation of digital images is not reflected in the
horizontal accuracy because the check points are measured
manually without any automatic or semiautomatic support.
Despite of the worse b/h factor a small improvement in the
vertical accuracy has been observed. This is due to the fact that
most of the check points are measured in images from different
strips (flown at 25% side lap), where the (larger) “side lap b/h
ratio” compensates the worse b/h in flight direction.
Automatically derived DEMs from digital and analog images
result in more or less comparable accuracies with small
advantages for the digital camera. Digital images also allow
smaller grid spacing down to the size of 5 pixels. The
comparison with L1DAR data proves a vertical accuracy of up
to /2 pixels in flat areas without vegetation and buildings. In
urban areas the accuracy is considerably worse (approximately
2 pixels), because the DEM grid points do not well represent
the height discontinuities at the edges of buildings or other man
made objects.
From the steroplotting point of view, the conclusion is that the
use of digital images allows digitizing more information,
extending the flight period and introducing more comfort in the
steroplotting process. Although there is a slight loss of the relief
feeling, and there is a small decrease in the height accuracy, it
is not important enough to introduce degradations in the final
products. Finally, more automatic tools are needed to optimize
image display and management. Stereoplotting check points
showed that an accuracy of 1.1 pixels in the horizontal
components and about 1.5 pixels in the vertical component can
also be routinely achieved in production projects.
Concerning the resolution of DMC digital images in the paper
absolute resolution measures have been generated from a
Siemens star target whose dimensions are known. The
measurement FWHM which is extracted from the LSF function
provides a specific quantification of the resolution of the system,
coherent with techniques of resolution evaluation, using USAF
targets. This very same methodology can also be applied to
analogues images (Alamus et. al, 2005). The results then
obtained, showed the greater resolution power of the DMC, in
front of the film cameras.
REFERENCES
Alamús, R., Komus, W., Palá, V., Pérez, F., Arbiol, R., Bonet,
R., Costa, J., Hernández, J., Marimon, J., Ortiz, M. Á., Palma,
E., Pla, M., Racero, S., Talaya, J. 2005: Validation process of
the ICC digital camera. ISPRS Worshop “High-Resolution
Earth Imaging for Geospatial Information”. Hannover.
Honkavaara, E., Jaakkola, J., Markelin, L., Becker, S., 2006.
Evaluation of Resolving Power and MTF of DMC. In
proceedings of ISPRS Comission I Symposium, Paris.
ACKNOWLEDGEMENTS
We would like to thank the hard job done by M. Cabré, J.
Costa., J. Marimon and A. Serrano for the selection and
measurement of the check points used in the aerial triangulation
and stereoplotting accuracy assessment.
1188
Ci
•ÂC/WV’ ;