International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
can be done in a number of ways; for instance, using a Bayesian
approach (Z.Q. Liu, 1997), Dumpsters — Shater (G. Shater,
1976), or fuzzy logic (J.C. Bezdek , 1999.). One technique that
has enjoyed success in other vision applications is the use of
fuzzy measures. In our approach, we follow this method.
We first treat each property (information) as fuzzy variables
and assign fuzzy density to all possible values. Fig.3 (a) is for
property $1 , which is the minimum distance from the average
color of the region to the prior-colors selected empirically,
which most of building roofs would probably present. There is a
reasonable hypothesis that most of build roofs are created with
materials of limit categories, so they have limit kinds of color
and texture. Fig.3 (b) and Fig.3 (c) are for properties #2 and
$3 , which are the high mean absolute deviation and low mean
absolute deviation of the histogram of curvatures. Fig.3 (d) is
for property 94, which is the ratio of width over height of the
region. We assert the two thresholds are 0.19 and 0.45
respectively in our test. Fig.3 (e) is for property $5 , Which is
the size of the rectangle. Most of the properties’ thresholds
except $4 are adapted to the images to be processed;
according to our experience, they are relatively common to the
same batch of images.
As a simple fuzzy integral, last decision is made by naively
computing the average of the fuzzy density to all possible
properties, and assert a threshold to determined "it is a building"
or "it is not a building".
a
Figure 4. One of tests in the images of Beijing, China
6. CONCLUSIONS
As shown in Fig.4, we made some experiments on some color
high-resolution images; it is one of the parts of an image of
Beijing area. The red line rectangles are the parts identified as
buildings by our algorithm. Verified in the available image
samples, we achieve a right identify ratio lager than 75%.
References from Journals:
Eric N. Mortensen, 2001. A Confidence Measure for Boundary
Detection and Object Selection. Proceedings of the 2001 1EEE
620
Computer Society Conference on Computer Vision and Pattern
Recognition.
Fua P.,1996. Model-Based Optimization: Accurate and
Consistent Site Modeling. Proceedings of the 18th SPRS
Congress, Comm. III, WG 2, Vienna, Austria, pp. 222-233.
Henricsson O., F. Bignone, W. Willuhn, F. Ade, O. Kubler, E.
Baltsavias, S. Mason, A. Grun, 1996. Project AMOBE:
Strategies, Current Status and Future Work. Proceedings of the
18^ SPRS Congress, Comm. IN, WG 2, Vienna, Austria, pp.
321-330.
Mark C.K. Yang, Jong-Sen Lee, 1997. Hough Transform
Modified by Line Connectivity and Line Thickness. /eee
Transactions On Pattern Analysis And Machine Intelligence,
Vol. 19, No. 8, pp. 905-910
S. Cohen and L. Guibas, 1999, The Farth Mover's Distance
under Transformation Sets.Proc. 7th IEEE Intel. Conf. Computer
Vision.
Weidner U.,1996. An Approach to Building Extraction from
Digital Surface Models. Proceedings of the 18th SPRS
Congress, Comm. III, WG 2, Vienna, Austria, pp. 924-929,
Z.Q. Liu, 1997. Bayesian paradigms in image processing. /nt. J
Pattern Recognition Artif. Intell. 11 (1 ),pp.3—34.
References from Books:
G. Shafer, 1976. A Mathematical Theory of Evidence, Princeton
University Press, Princeton, NJ.
G.Wyszecki., 1982. Color Science: Concepts and Methods,
Quantitative Data and Formulae. John Wiley and Sons, New
York, NY.
J.C. Bezdek, J.M. Keller, R. Krishnamupram, S.K. Pal, 1999.
Fuzzy Models and Algorithms for Pattern Recognition and
Image Processing. Kluwer Academic Publishers, Norwell, MA.
Wilhelm Klinggenberg, 1978. A Course In Differential
Geometry. Springer-Verlag.pp.3.
References from Other Literature:
Grun A., O. Kubler, P. Agouris, 1995. Automatic Extraction of
Man-Made Objects from Aerial and Space Images, Virkhauser
Verlag, Basel, pp. 199-210.
Mark A. Ruzon, 1999. Color Edge Detection with the Compass
Operator. Computer Science Department Stanford University
Stanford, CA 94305.
Robert W. Carroll. Report of Detecting Building Changes
Through Imagery And Automatic Feature Processing. Hitachi
Software Global Technology, Ltd.10355 Westmoor Drive, Suite
250 Westminster, Colorado 80021 303-466-9255,
Rcarroll@HSGT.com
References from websites:
Monge Patch, Mathworld, USA ;
http://mathworld.wolfram.com/MongePatch.html (accessed 22
Apr .2004)
KEY M
ABSTR
The hig
professi
Informa
will pro
simple :
satellite
primitiv
Our har
and offl
Softwar
componi
manager
function
has the í
for Lanc
service,
want to
which h
images.
Sing
multispe
environn
map prox
establish
acquisiti
between
makes th
of natior
images +
national
museum.
Tho
body anc
and do it
Korea, i
overlapp
processir
because
technolo;
each yea
establish
of satelli
the instit
e
Corre:
