The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
1127
: (a) After segmentation
(b) Merged segmentation
(c) After thresholding
(d) Classification result
Figure 5 Result of each processing step
As explained in previous section, classification method
considering the spectral trait and object-based is appropriate for
classification technique of high resolution image and general
process is as Picture 4. Since capacity of image for total target
region is large and considerable processing time is consumed,
test was carried out by sampling the housing concentrated area.
In this process, determination on factor for each classification
stage is attempted.
It was classified that 8.5% was residential read and road, 76.7%
was forest and cultivate site and 14.8% was water. Since it was
filmed in October, agricultural products were abundant in
overall cultivate site but it could be classified together in aspect
of forest and vegetation index thus it was harvesting period. It
was classified as 32.2% of cultivate site and 44.5% of forest.
3.4 Land Used Map
Figure 5(a) presents the test sampling region for coverage
classification among target region and (b) is the image
segmented by setting proper scale. As a result, road, building,
forest, etc were detected. It is the image displayed with average
value of segments after combining adjacent segments,
Figure5(c) presents the allotted value for each segment of
sample region and final classification image was sought after
granting the color to each ROI after determining the threshold
so that segments can be classified well.
Figure 6 Final result
Land use
Ratio I
residential area +road
8.5%
cultivate site
32.2%
forest
44.5%
water
14.8%
Land category was extracted and symbol color was granted for
each land use using ArcGIS Desktop 9.2 from annotation
3<Nul>
O historical places
□ orchard
□drain
■ needy
■ habitation
■ road
□ pastire
□ grave
□fish terms
□ swamp
■forest
□ mixed use
£3 industrial a-ea
□ crop iekJ
□ empanlcment
□ reSgicn area
□ gas station
□pairing lots
□ warehouse
■ rivers
□ school zone
Figure 7 Land use map
among layer of 1:1000 cadastral map. Land use map was
created as Picture 11 classifying 21 land categories and non-
classified as null. Grouping was carried out to compare the land
coverage classification using image and it was combined as 5
classes as Diagram 5.
Land use
Parcels
area()
Ratio II
residential area
4289
4482367
6.2%
cultivate site
15679
24301324
33.4%
water
2558
11440853
15.7%
forest
4757
29880693
41.0%
road
3549
2460073
3.4%
null
153
245956.6
0.3%
30985
72811267
100%
Table 3 The number of parcel and area with respect to class
Table 2 Classification ratio
Ratio of each class was acquired as residential area: 6.2%,
cultivate site: 33.4%, water: 15.7%, forest: 41%, road: 3.4%
and null: 0.3% from total region.