is required and enters the code of the data layers to be
queried. After a search through the database, the
attributes found for that location are displayed. The
polygons surrounding this location in each layer may
also be displayed if so requested.
An alternate version of this procedure is to localize the
query by delineating a window in the image. The
database search then returns the attributes of all
polygons which are inside or intersected by this
window and the polygons in question are displayed.
Another situation that could arise in spatial analysis is
that new information is sought within certain
geographic boundaries or the current information has to
be verified or updated. These boundaries are either
stored already in the database or may be delineated by
the analyst. Digital image classification techniques can
then be employed for this purpose. A prerequisite is,
however, that suitable training samples can be located
for each information class searched for.
Two procedures can be followed in this type of query
[Derenyi and Turker, 1996]. In the first case, a per-
pixel image classification is performed covering the
entire study area, using one of the standard algorithms
such as the parallelepiped or the maximum likelihood
classifier. Thereafter, a polygon-by-polygon assessment
of the result is performed to establish the class
distribution within the specified geographic boundaries.
The individual polygons are located through the DBMS
for assessment.
In the second case, the image classification is bypassed
entirely and only image statistics are generated within
the polygons to be analyzed. Mean, standard deviation
and median values are possible statistics selected. A
comparison of these statistics with those obtained for
the training samples forms the class assignment of each
polygon. This approach is especially suitable for
monitoring and change detection.
Database queries may be issued through any kind of
imagery, be it airborne or spaceborne, monochrome,
colour or multispectral. The image must of course be
geometrically registered to the geographic database.
Image query through classification is primary intended
for spaceborne multispectral imagery. Decision
making based on image statistics only is especially
efficient for analyzing single or small groups of
polygons scattered through a study area. The polygons
do not have to be pre-established but can be drawn by
the analyst in an interactive mode. All information
generated through the image supported spatial analysis
can, of course, immediately move into the database.
This spatial analysis scheme was successfully
implemented and tested in the Computer Aided
Resource Information System (CARIS), GIS which
supports both vector and raster data handling [Derenyi
and Pollock, 1990; Derenyi, 1991].
4. CONCLUSIONS
Incorporating images within a GIS for spatial analysis
218
is a powerful tool with several advantages over the
conventional, vector data only, technique.
* Image analysis can be performed within known
geographic boundaries, which are stored in the GIS
and can be queried through the DBMS.
* Database queries can be issued through image
displays which provide a more complete and detailed
view of the real world than graphical displays do.
* The new information generated through digital
image analysis can directly be entered into the
database.
* The interactive query location selection feature of
this scheme is especially attractive for exploratory,
browsing type spatial analysis.
ACKNOWLEDGMENT
The authors wish to acknowledge the contribution of
Mustafa Türker who has implemented some of the
concepts presented in this paper.
REFERENCES
Derenyi, E. and R. Pollock, 1990. Extending a GIS
to support image-based map revision.
Photogrammetric Engineering and Remote Sensing,
Vol. 56, No. 11, pp. 1493-1496.
Derenyi, E., 1991. Design and development of a
heterogeneous GIS. CISM Journal ACSGC, Vol.
45, No. 4, pp. 561-567.
Derenyi, E. and M. Turker, 1996. Polygon based
image analysis in a GIS. Proceedings of the 19th
Canadian Symposium of Remote Sensing,
Vancouver, British Columbia.
Edwards, G., 1993. The integration of remote sensing
and GIS: Fundamental questions and new
approaches. Proceedings of the 16th Canadian
Symposium of Remote Sensing, Sherbrooke,
Quebec, pp. 873-878.
Petersen, P., 1900. Living Proof. Colorado Springs,
U.S.A., NAVPRESS.
Price, K., 1991. "Spatial modeling using geographical
information systems." The ABC's of GIS.
American Society of Photogrammetry and Remote
Sensing. :
Pries, R., 1995. A system for large scale image
mapping and GIS data collection. Photogrammetric
Engineering and Remote Sensing, Vol. 61, No. 5,
pp. 503-511.
Trotter, C.M., 1991. Remotely sensed data as an
information source for geographical information
systems in natural resource management: A review.
International Journal of Geographical Information
Systems, Vol. 5, No. 2, pp. 225-239.
Unwin, D., 1981. Introductory Spatial Analysis. New
York, U.S.A., Metheun and Co. Ltd.
Wilson, J., 1995. Client-vendor partnership advances
digital aerial photography system. GIS World, Vol.
8, No. 9, pp. 44-47.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
KE
The
tecl
pro
req
In 1
rea
The
phc
um
At 1
SOU
bac
up
Lar
tim
bac
tecl
In t
Ter
rea
(FC
(MF
anc
anc
Aft
inte
apr
The
con
Sys
diff
pro
enc
mo
tele
sur
pro
The
infc
tas
con
infc
Co
-C
-P
-W
-F
