The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008
simultaneously. The epipolar line constraints are also used
indirectly in the model.
Figure 4 Initial DSM from matching points
solution and occlusion problems effectively. Moreover, dense
ground points and DSM can be generated through the
combining of the algorithm model and some quality control
measures.
However, massive ground points are not enough to generate
precise DSM from high resolution images for urban areas.
During the process of grid DSM interpolation, discontinuities
are smoothed. This will lose a lot of information such as
buildings’ edge and make parts of DSM connected, which can
be seen from Fig 4 and Fig 5. In order to generate more precise
and practical DSM, edge features have to be extracted, matched
and combined with point features in DSM generation, which
will be our work in the future.
REFERENCES
X.H. Xiong, Y. Chen, and Z.B. Qian, A Fast, Accurate and
Robust Image Matching Algorithm, Acta Geodaetica
Cartographica Sinica, 2005, 34(1): pp. 40-44.
Y. S. Zhang, D.Z. Fan, and S. Ji, Multi-view Matching
Algorithm Model for ADS40 Sensor, Journal of Zhengzhou
Institute of Surveying and Mapping, 2007, 24(2): pp. 83-86.
Z. X. Zhang, From Digital Photogrammetry Workstation (DPW)
to Digital Photogrammetry Grid (DPGird), Geomatics and
Information Science of Wuhan University, 2007, 32(7): pp.
565-571.
Okutomi, M., Kanade, T. A Multiple-baseline Stereo. PAMI,
1993, Vol. 15, No. 4, pp. 53-363
Y.J. Zhang, Image Interpretation and Computer Vision,
TSinghua University Press, Beijing, 1999. pp. 87-102.
D.Z. Fan. Theory and Algorithms of DSM Generation from
Multi-line-array Images Matching, Institute of Surveying and
Mapping for the degree of Doctor of Technical Sciences, 2007.
pp. 35-57.
Figure 5 Optimized DSM after quality control
From the experiments currently done, the algorithm shows great
ability to improve matching reliability and can solve multiple