International Archives of Photogrammetry and Remote Sensing, Vol. 32, Part 7-4-3 W6, Valladolid, Spain, 3-4 June, 1999
96
The examples shown and discussed have been selected for
demonstrating the principle of DTM integration based on
wavelets and give some first ideas about the accuracy behaviour
of the integration approach on different scales. But without
doubt, more comprehensive investigations have to follow to
support these first results.
5. CONCLUSIONS
In this paper, we presented a concept for the integration of
DTMs which differ in scale. The basic idea of the concept is to
use wavelets for representing DTMs on multiple scales. The
generation of a multiscale representation is obtained by wavelet
decomposition of the high resolution DTM. For merging two
DTMs, the DTM representations on the same scale or resolution
level are introduced into a weighted least squares fit. With
wavelet reconstruction, the integrated DTM estimates on all
finer scales are obtained.
Some examples are investigated to demonstrate the applicability
of the proposed wavelet approach and the first experiments with
the developed routines show that the errors of the fused
estimates at the fine scale have been reduced. This basically
confirms our a priori expectation for the integration result.
Further investigations have to follow. In particular, DTMs with
given ground truth should be taken into account to rigorously
assess the overall accuracy improvement. Comparison with
other methods would be also quite interesting. Further, little
attention has been paid to the wavelet itself. The standard
wavelets, like Haar and Daubechies, are only two among a lot of
others, even though they excel because they are simple,
orthogonal, have compact support, certain smoothness and other
beneficial properties.
Finally, the merging by simple least squares weighting might be
replaced by more sophisticated DTM interpolation schemes.
Even though these aspects have no impact on the basic concept,
they have to be taken into account in future experimental
investigations.
REFERENCES
Abidi, M.A. and R.C. Gonzalez, 1992. Data Fusion in Robotics
and Machine Intelligence. Academic Press, Boston, 546 p.
Ackermann, F. and M. Hahn, 1991. Image Pyramids for Digital
Photogrammetry. In: Digital Photogrammetric Systems,
Wichmann Verlag, Karlsruhe, pp. 43-58.
Daubechies, I., 1992. Ten Lectures on Wavelets. SIAM
Publishers, Philadelphia.
Li, Z., 1996. A Framework for Multiresolution Approximation
and its Applications to Physical Geodesy. Dissertation,
Department of Geomatics Engineering, The University of
Calgary.
Louis, A.K., MaaB, P. and A. Rieder, 1997. Wavelets: Theory
and Applications. Wiley & Sons, Chichester, 324 p.
Mallat, S., 1989. A theory for multiresolution signal
decomposition: the wavelet representation. IEEE Transactions
on Pattern Analysis and Machine Intelligence, Vol. 11, No. 7,
pp.674-693.
Pan, Y. and B. Schaffrin, 1999. Bayesian Wavelet Analysis for
Image De-Noising - The First Steps. Photogrammetrie,
Fernerkundung, Geoinformation, Vol. 2, pp. 95-100.
Thierschmann, M., Martin, U. and R. Rosel, 1997. New Per
spectives in Image Compression. In: Photogrammetric Week
'97, D. Fritsch/D. Hobbie (Eds.), Wichmann Verlag,
Heidelberg, pp. 189 - 199.