An object-oriented pattern recognition approach for urban classification - Urban Remote Sensing: Monitoring, Synthesis and Modeling in the Urban Environment - page 132

7 Refine the rule set, if necessary.
8 Reclassify (if not unsatisfactory, develop a new strategy).
9 Apply the best rule set to the entire dataset.
10 Classify the entire image.
11 Export results.

Congalton, R.G. (1991) A review of assessing the accuracy of
classifications of remotely sensed data. Remote Sensing of
Environment , 37 , 35-46.
Definiens. 2007. Definiens Developer 7.0, Reference Book ,
Munchen, Germany, p. 195.
Definiens. 2008. Definiens Developer 7.0, User Guide ,Munchen,
Germany, p. 506.
Grimmond, C.S.B. and Oke, T.R. (2002) Turbulent heat fluxes
in urban areas: Observations and local-scale urban meteoro-
logical parameterization scheme (LUMPS). Journal of Applied
Meteorology , 41 , 792-810.
Im, J., Jensen, J.R., and Tullis, J.A. (2008) Object-based change
detection using correlation image analysis and image segmen-
tation techniques. International Journal of Remote Sensing , 29 ,
399-423.
Ivits, E. and Koch, B. (2002) Object-oriented remote sensing
tools for biodiversity assessment: a european approach, in
Proceedings of the 22nd EARSeL Symposium, Prague, Czech
Republic, 4-6 June 2002 , Millpress Science Publishers, Rotter-
dam, Netherlands.
Ketting R.L. and Landgrebe, D.A. (1976) Classification of multi-
spectral image data by extraction and classification of homo-
geneous objects. IEEE Transactions on Geoscience Electronics
GE , 14 , 19-26.
Lee, J.Y. andWarner, T.A. (2006) Segment based image classifica-
tion. International Journal of Remote Sensing , 27 , 3403-3412.
Lillesand, T.M., Kiefer, R.W. and Chipman, J.W., (2008) Remote
Sensing and Image Interpretation , 6th edition, John Wiley &
Sons, New York.
Lizarazo, I. and Barros, J. (2010) Fuzzy image segmentation for
urban land-cover classification. Photogrammetric Engineering
& Remote Sensing , 76 , 151-162.
Myint, S.W., Giri, C.P., Wang, L., Zhu, Z. and Gillette, S. (2008).
Identifying mangrove species and their surrounding land use
and land cover classes using an object oriented approach with
a lacunarity spatial measure. GIScience and Remote Sensing ,
45 , 188-208.
Navulur, K. (2007) Multispectral image analysis using the object-
oriented paradigm . CRC Press, Taylor & Francis Group, Boca
Raton, FL.
Pal, R. and Pal, K. (1993) A review on image segmentation
techniques. Pattern Recognition , 26 , 1277-1294.
Stow, D., Lopez, A., Lippitt, C., Hinton, S. and Weeks, J. (2007)
Object-based classification of residential land use withinAccra,
Ghana based on QuickBird satellite data. International Journal
of Remote Sensing , 28 , 5167-5173.
Strahler A.H., Woodcock C.E. and Smith, J.A. (1986) On the
nature of models in remote sensing. Remote Sensing of Envi-
ronment , 20 , 121-139.
Walker, J.S. and Blaschke, T. (2008) Object-based land-cover
classification for the Phoenix metropolitan area: optimization
vs. transportability. International Journal of Remote Sensing ,
29 , 2021-2040.
Yu,Q.,Gong,P.,Clinton,N.,Biging,G.,Kelly,M.andShirokauer,
M. (2006) Object-based detailed vegetation classification with
airborne fine spatial resolution remote sensing imagery. Pho-
togrammetric Engineering & Remote Sensing , 72 , 799-811.

The same strategy should also be considered for the clas-
sification with nearest neighbor classifier. However, the results
obtained from testing data (small subsets) should be consid-
ered to provide general guidance for implementing the nearest
neighbor classifier. One still needs to select object training sam-
ples when using the entire dataset. The classification accuracy
is greatly influenced by the selection of training samples. It is
important that one selects as many spectrally distinct training
objectsaspossiblewhenclassifyinganentirestudyarea.
Our experience with Definiens/eCognition software package
is that some objects that appear visually similar and belong to
two different classes may still be identified separately and accu-
rately by selecting many different types of training samples from
within problem areas. This may stem for the Definiens classifiers
being based on non-parametric rules that are independent of a
normal distribution. The object-oriented approach allows addi-
tional selection or modification of new training samples (training
objects) after multiple iterations of a nearest neighbor classifi-
cation, until a satisfactory result is obtained. There are many
possible combinations of functions, parameters, object scale lev-
els, features, and variables available with the object-oriented
approach. The successful use of nearest neighbor classifier in
the object-based paradigm largely relies on repeatedly modifying
training objects as a trial-and-error approach. Nonetheless, the
object-based classification approach seems to be very effective at
classifying urban LULC categories, especially when based on fine
spatial resolution multispectral imagery.

References

Baatz, M. andA. Schape. (1999) Object-Oriented andMulti-Scale
Image Analysis in Semantic Networks. In Proceedings of the
2nd International Symposium on Operationalization of Remote
Sensing, 16-20 August 1999 . Enschede, ITC.
Baatz, M. and Schape, A. (2000) Multiresolution segmentation-
an optimization approach for high quality multi-scale image
segmentation, in Angewandte Geographische Informationsver-
arbeitung (eds XJ. Strobl, T. Blaschke and G. Griesebner),
Wichmann-Verlag, Heidelberg, pp. 12-23.
Benz, U.C., Hofmann, P., Willhauck, G., Lingenfelder, I. and
Heynen, M. (2004) Multi-resolution, object-oriented fuzzy
analysis of remote sensing data for GIS-ready information.
ISPRS Journal of Photogrammetry and Remote Sensing , 58 ,
239-258.
Blaschke, T., Burnett, C. and Pekkarinen, A. (2004) New con-
textual approaches using image segmentation for object-based
classification, in Remote Sensing Image Analysis: Including the
Spatial Domain (eds F. de Meer and S. de Jong), Kluwer
Academic, Dordrecht, pp. 211-236.