oul 2004
——
measure
ypes) of
lity ofa
be then
rder to
ot only
also the
gned to
> shades
1ades of
entation
y of the
ties can
on help
at have
form in
'e from
S. ]t is
les are
liability
tour
-
VM
£2
| Table
bution
value
7.0.07
mpute
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
LANDSAT:
tx 7 0,1926 « 1,703 and rg — 0,3977 « 1,703 (there is
no trend in these directions)
CBERS:
tv = 0,2828 < 1,703 and tg 7 1,2172. « 1,703 (there is
no trend in these directions)
42 Accuracy Analysis
Accuracy analysis was carried out for both Landsat and CBERS
images and the results of the estimates of Y nd and Pints
computed using Equation (3) and (4), the first part of Table 3
and Table 1, are presented in Table 4.
CLASSES | DIRECTIONS |LANDSAT| CBERS
A E 1.265.172| 1.317.865
N 1.387.768| 1.322.864
B E 455.463 | 474.432
N 499.598 | 476.232
C E 316.293 | 329.466
N 346.942 | 330.716
Table 4. Estimated values of 7^4, and nl used in the
accuracy analysis for the Landsat and CBERS images.
Adopting the critical value of 7? — 36,7412 and considering the
limit values for the classes A, B, and C as shown in Table 1 for
the Brazilian Map Accuracy Standards (scale 1:100.000), these
two images have accuracy equivalent to the class C.
43 Generated Point Method
The values of the Standard Error (SE) for the generated point
method are the same as the Brazilian Map Accuracy Standards,
for instance, (see Table 1). As this method works with relative
distances between homologous points, there is no need to apply
trend analysis to check for the presence of systematic errors in
the E and N directions.
The.accuracy analysis was also performed for both Landsat and
CBERS images using a set of equally spaced homologous
coordinates in order to check the positional feature accuracy for
à given confidence level. The results are presented in Table 5.
CLASSES | LANDSAT | CBERS
A 128,0141 | 441,437
B 46,0852 158,918
C 32,0035 110,359
Table 5. Estimated values of y^p, used in the accuracy
analysis for the Landsat and CBERS images
The critical value of 7^ is 48.2329 and the limit values are given
In Table 1 for the Brazilian Map Accuracy Standards (scale
1:100.000). The Landsat image has an accuracy equivalent to
class B, while the CBERS product does not fit in any specify
class.
S. CONCLUSIONS
Trend analysis indicates that there is no systematic error in any
direction for the Landsat and CBERS geometrically corrected
Images at the 9094 confidence level. On the other hand, both
983
corrected image have an accuracy equivalent to class C using
the Brazilian Map Accuracy Standards for a map scale of
1:100.000. Only the Landsat image fits the class B
specification, using the generated point method.
These results show that a considerable amount of research
needs to be undertaken before the spatial characterization of
positional and thematic accuracy associated with remote
sensing data can be adequately reported in standardized format
and legends. Several techniques for the quality control of spatial
databases using generic features are adapted to the context of
remote sensing. However, one drawback for some of these
approaches is the difficulty in obtaining homologous points in
both representations. Alternative techniques for overcoming
such limitations could be the use of a spline Fitted method, in
which the shape of curves fitted using splines are compared
instead of isolated homologous control points (Galo et al.,
2001).
6. ACKNOWLEDGEMENTS
Preliminary research by Dr. Vieira was supported by the
Brazilian Research Council (CAPES). The later stages of this
study were conducted as part of the CNPq/COAIE/Kit Enxoval
(Project Number 68.0045/01-5) performed in collaboration with
The University of Nottingham. We are grateful to Logica PLC,
INPE and SPOT Image for permission to use their images.
7. REFERENCES
Arbia, G., Griffith, D., and Haining, R., 1998. Error
propagation modelling in raster GIS: overlay operations.
International Journal of Geographical Information Science, 12,
pp. 145-167.
Congalton, R. G., and Green, K., 1999. Assessing the Accuracy
of Remotely Sensed Data: Principles and Practices. Lewis
Publishers, New York.
Ferreira, L. F., and Cintra, J. P., 1999. Quantificaçäo de
Discrepáncias entre Feigóes Lineares por Retángulo
Equivalente. Brazilian Magazine of Cartography, 51, pp. 1-8.
Galo, M., Dal Poz, A. P., and Ferreira, F. M., 2001. Uso de
feiçôes no controle de qualidade em cartografia. In Proceedings
of the XX CBC - Cartography Brazilian Congress, Porto
Alegre, 7 - 12 October 2001,
http://www.cartografia.org.br/200 l/english/index.html (acessed
10th Oct. 2001).
Jassen, L. L. F., and van der Wel, F. J. M., 1994. Accuracy
assessment of satellite derived land-cover data: a review.
Photogrammetric Engineering and Remote Sensing, 48, pp.
595-604.
Jensen, J. R., 1986, Introductory Digital Image Processing — A
Remote Sensing Perspective. Englewood Cliffs, NJ, Prentice-
Hall.
Lunetta, R. S., Congalton, R. G., Fenstermaker, L. K., Jessen, J.
H., and Mcgwire, K. C., 1991, Remote sensing and geographic
information system data integration: error sources and research
issues. Photogrammetric Engineering and Remote Sensing, 57,
pp. 677-687.
Merchant, D. C., 1982. Spatial accuracy standards for large
scale line maps. In Proceedings of the Technical Congress on
Surveying and Mapping, 1, pp. 222-231.
