International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
(c)
Figure 2. Image fusion of QuickBird images for object
extraction. (a) Original Pan image; (b) Original MS image; (c)
Pan-sharpened image.
2.2 Classification
To extract road networks according to their spectral information
and to get an acceptable result, the original QuickBird MS
image and the pan-sharpened MS image were classified in this
study using multispectral classification. Comparisons showed
that the classification results from pan-sharpened MS images
are significantly better than those from original MS images. The
reason for this improvement is that the pan-sharpened image has
higher resolution, and the spectral information is well reserved
from the original MS image. Therefore, the pan-sharpened
image was chosen for the road classification.
Since an unsupervised clustering method is usually better
suited for classifying heterogeneous classes in high resolution
satellite images than a supervised classification (Jensen et al.
1994, Csatho and Schenk 1998, Zhang 2001), the unsupervised
fuzzy K mean clustering method was used to classify the
QuickBird images in this study.
Figure 3a shows the clustering result from the pan-sharpened
QuickBird image in an urban area. Figure 3b is the classified
binary road image from the clustering result (Figure 3a). It is
clear that almost all the road networks are correctly extracted.
However, the rate of misclassification is high. For example,
there are small family driveways connected to road networks,
and many house roofs are classified into the road networks.
These make it impossible to obtain an accurate road network
without further processing.
~ s. xo HS aam
(b)
Figure 3. The unsupervised road classification of pan-
sharpened QuickBird MS image. (a) A close-up of the
clustering result in an urban area. (b) The classified road image
after elimination of speckles.
Sob
Rol
ima
Ho
det
nee
Gal
blu
con
