1182
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008
Figure 12 A point to node comparison highlighting change
between 2005 and 2006 Bournemouth datasets.
4. DISCUSSION
The use of point cloud data derived from aerial photogrammetry
has met with varying success in both of the test sites presented.
A number of spurious points are visible in the results of each
comparison. Mainly along significant edges in each scene, such
as buildings and vegetation. Slight variations in the point clouds,
perhaps arising from poor image correlation during surface
extraction, led to subtle variations in the structure of the octree
and thus in the changes identified. Especially along edges where
image correlation may be poor. The quality of the input data is
of prime importance when attempting to identify small changes.
While it would be preferable to not have to edit digital surface
models derived from imagery, in order to improve the success
of the results editing of the surface models is required.
The proposed approach is perhaps better suited to datasets that
are collected from a different perspective and with a larger
number of points per m 2 with respect to the level of change
anticipated. For example, this approach would be of interest in
comparing datasets from ground based platforms, such as data
collected by the StreetMapper mobile mapping system (Barber
et al, 2008).
In this study, the approach has been to take a generic approach
to the problem and identify differences between two sets of
point cloud data, so a user could investigate any of the changes
identified. If a user was only interested in mapping revision a
more appropriate approach might be to compare newly collected
data with the mapping database directly. However, by taking a
generic approach to the problem the proposed solution may be
of interest to a wider range of users. For example determining
change between datasets where detailed existing mapping does
not exist such as in aerial reconnaissance or in a situation where
near real-time change detection is required.
This study has presented results from an urban/peri-urban area.
The limiting factor in applying this type of algorithm
operationally would be the frequency with which data could be
collected. If, as for the continuous revision process, change
needs to be recorded within six months of it occurring, data
would need to be collected more frequently than this. While, as
discussed earlier, this may be possible with increasingly more
efficient data capture flow lines, at present this is unrealistic, so
such an approach would only be of use for the cyclic revision
process that requires data less frequently.
Further work includes the filtering of results so as to more
clearly differentiate between significant changes, insignificant
changes (such as vegetation growth) and changes relating to
errors in the data collection process. Also, further testing is
required using point clouds collected by airborne lidar sensors.
5. SUMMARY AND CONCLUSIONS
The use of an octree data structure for the detection of change
between datasets has shown some limited success. Areas of real
world change have been identified from real datasets, but a
significant amount of spurious results have been generated,
possibly relating to the quality of the input datasets which were
generated from aerial photography. Improved results are
anticipated when lidar derived point clouds are used as input
data. Moreover, it is anticipated that the use of octree data
structures would facilitate the future integration of airborne and
ground based datasets.
ACKNOWLEDGMENTS
This investigation was conducted by Newcastle University
whilst Dr Barber was a researcher in the School of Civil
Engineering and Geosciences. The research was funded and
supported by Ordnance Survey. BAE Systems kindly provided a
licence for the NGATE module for use in the project.
REFERENCES
Barber, D. M., Mills, J. P. and Smith-Voysey, S. 2008.
Geometric validation of a ground-based mobile laser scanning
system. ISPRS Journal of Photogrammetry and Remote Sensing.
63(1): 128-141.
Besl, J. P. and McKay, N. D., 1992. A method for registration
of 3-D shapes. IEEE Transactions on Pattern Analysis and
Machine Intelligence, 14(2): 239-256.
Botsch, M. Wirantanya, A. and Kobbelt, L., 2002. Efficient
high quality rendering of point sampled geometry. Proceedings
of Eurographics Workshop on Rendering.
Girardeau-Montaut, D., Roux, M., Marc, R. and Thibault, G.,
2005. Change detection on point cloud data acquired with a
ground based laser scanner. The International Archives of
Photogrammetry, Remote Sensing and Spatial Information
Sciences, 36(3/W 19).
Holland, D. 2008. Automating topographic change detection at
Ordnance Survey. Geomatics World. 16(3) 30 - 32.
Ordnance Survey, 2006. Revision policy for basic scale projects.
Ordnance Survey Information Paper. 3 pages.