The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
executing rule 6 executing rule 5
Figure 7. Soft-decision change detection based on rules
By applying the same accuracy analysis method applied to the
hard-decision change detection result, from Table 8, it can be
seen that the soft-decision analysis can generate improved
change detection result, the omission error is 6.5%, the
commission error is 15.3%; compared with Table 6, the overall
accuracy is improved from 62.7% to 78.2%, and the
commission error is reduced by half.
Reference data
Sum in row
Classification
data
unchanged
changed
unchanged
161
39
200
changed
92
308
400
Sum in column
253
347
600
Table 8. Accuracy evaluation for the soft-decision change
detection
'^^2004
2005^"\
vegetation
built-up land
bare land
water body
vegetation
0.0
0.0
0.39
built-up land
149.0
7.5
4.8
bare land
173.6
0.0
18.8
water body
7.8
0.0
0.0
Table 9. Area change of the land use types of the test site (Area
unit: hectare)
The statistic of the change of the area of each land use type in
the research area (Table 9) shows that from October 2004 to
October 2005, the main land use/cover changes are the decrease
of vegetation, as well as the increase of the built-up land and
the bare land. By further analyzing the classified result of the
SPOT5 image acquired in 2004, we find out that in the
trajectory of vegetation to built-up land, 20% of vegetation is
cropland and 80% of vegetation is grassland. Because the study
area is located at the downstream region of Haihe River, which
belongs to the fluvial plain and marine plain, the annual mean
relative humidity is high, and the case of vegetation changed to
bare land usually occur when the parcel has been authorized to
be used as built-up land but construction has not been done or
just at the beginning of the construction. The circumstance of
the decrease of vegetation, as well as the increase of the built-
up land and the bare land well reflects the phenomenon and the
trend of urban expansion and the occupation of the cropland
and the unused land at the urban fringe areas.
4. CONCLUSIONS
Because of the different imaging mechanisms, optical imagery
and SAR data has different and complementary image
characteristics and information content. In order to overcome
the insufficiency of single remote sensing data source during
change detection, and to make use of the complementary
characteristics of these two kinds of data source, this paper
develops the idea and presents the approach to land use/cover
change detection by different temporal SAR and optical image
integration. Data adopted in this research are SPOT5 image and
Radarsat-1 SAR data. With the successful launch of the high-
resolution and polarized SAR data since the end of last year,
such as Cosmo-SkyMed (Italy), TerraSAR-X (German) and
Radarsat-2 (Canada), it provides us more choices for SAR data
selection. With the abundant textural information and rich
polarization information presented in these data, change
detection accuracy will be improved considerablely.
REFERENCES
Baatz M and Schape A. 1999.Object-oriented and multi-scale
image analysis in semantic networks. Proceedings of the 2nd
international symposium on operationalization of remote
sensing.
Benz U C, et al. 2004. Multi-resolution, object-oriented fuzzy
analysis of remote sensing data for gis-ready information.
ISPRS Journal of Photogrammetry and Remote Sensing, No.58,
pp.239-258.
Blaschke T, et al., 2000.Object-oriented image processing in an
integrated GIS/remote sensing environment and perspectives for
environmental applications. Environmental Information for
Planning, Metropilis-Verlag, pp.555-570.
DE Bruin S. and Gorte B. G. H. 2000. Probabilistic image
classification using geological map units applied to land-cover
change detection. International Journal of Remote
Sensing,Vol.21, No. 12, pp.2389-2402.
eCognition 4.0 User Guide.2002. Defmiens Image GmbH.
German.
Guo Huadong. et al., 2000. Theories and applications of Radar
systems for Earth observation. Beijing: Science Publisher.
Siamak Khorram et al.1999. Accuracy assessment of remote
sensing-derived change detection. Published by American
Socitety for Phorogrammetry and Remote Sensing, Maryland,
USA.
Shi W Z and Ehlers M.1996. Determining uncertainties and
their propagation in dynamic change detection based on
classified remotely-sensed images. International Journal of
Remote Sensing, Vol. 17, No. 14, pp.2729-2741.
Van der Wei, F. J. M., Van der Gaag, L. C., and Gorte, B. G. H.,
1998. Visual exploration of uncertainty in remote sensing
classification. Computers and Geosciences, 24, pp.335-343.
1082