Transcript Comparison of classical classifies and object

Comparison between pixel-based
and object-oriented classification
approaches in urban area of the
arid environment
Qian Jinga,b, Zhou Qiminga, Hou Quana
a Department of Geography, Hong Kong Baptist University, Kowloon Tong,
Kowloon, Hong Kong
b Xinjiang Institute of Ecology and Geography, Chinese Academy of
Sciences
Outline
Introduction
Study area and data
Methods
Results and discussion
Conclusion
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
2
Introduction
 Urban development is one of the major forces
causing environmental change in aridzone of
China.
 In Xinjiang, expansion of urban areas is
concentrated within limited space of oases, with
constraints such as water resources.
 With the rapid increasing population, the
expansion of built-up areas are accelerated in
the past decades.
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
3
Introduction
 Delineating built-up areas from its background
has been a constant challenge in remote
sensing image processing (Erbek et al., 2004;
Lo and Choi, 2004).
 With the increasing availability of high-resolution
imagery, research has been focused on
automated delineation of built-up areas using
the images that have high frequency spatial
variance with limited spectral resolution.
 Object-oriented approach has been developed
for the segmentation of images for this.
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
4
The issues addressed
 In aridzone of China, the built-up areas are often
surrounded by farmland.
 However, they may also confuse with nearby
bare soil and stony desert, which present very
similar spectral characteristics as construction
materials such as concrete.
 The traditional pixel-based classification typically
yield large uncertainty in the classification
results.
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
5
Features in
arid area
Built-up areas
represented on the
Landsat ETM+ image
show different types of
cities with significant
spatial and spectral
variations on the
images. Also notice
the spectral similarity
between bare ground
(river bed) and small
cities and settlements.
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
6
Objectives
 To find the most appropriate approach for autoclassification of built-up areas for the aridzone of
China.
 Constraints:
Large areas to be covered in a short period time.
Limited availability of high-resolution images.
Cost-benefit concern
 Comparison between different classification
approaches is addressed by this paper.
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
7
Study area and data
Landsat ETM+ image of the study area, acquired on 7
August 2000).
2015/7/18
Map of the study area: the Centre Town of Manas County,
City of Shihezi and part of regimental farm of Division 8, at
North Xinjiang Economic Zone, China
International Conference on China's Urban Land and Housing in the 21 Century
8
Study area and data
 Centered at the city of Shihezi at north slope of
Tianshan Mountain, Xinjiang Uygur Autonomous
Region of China.
 At the centre of a large oasis.
 Three types of cities and settlements are
identified:
Shihezi: major city in the region, population ~200,000
Manas: county centre, population ~20,000
Liangzhouhu town: settlement, population ~ 5000
 Landsat ETM+ image is used, acquired on 7
August 2000
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
9
Data pre-processing
Geometric correction: image-to-image
registration using a geo-coded SPOT Pan
image of 2002 as master image, with 37
Ground Control Points. The RMSE is less
than 0.5 pixels
Reference data: interpreted from aerial
photos acquired in 2000
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
10
Image classification approaches
Pixel-based
Object-oriented
Philosophy
Pixel as element
Object as element
Information used
Spectral only
Spectral and spatial
Performance
2015/7/18
Expected to be superior, due
to additional spatial
information used
International Conference on China's Urban Land and Housing in the 21 Century
11
Methods tested
Normalized Difference Built-up Index
(NDBI)
Maximum Likelihood Classifier (MLC)
Object-oriented (O-O) image analysis
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
12
NDBI
 A pixel-based approach
 Using similar concept of vegetation index by
delineating built-up areas from other background
categories, only two bands of spectral data are
used.
 The test classification attempts to delineate builtup areas & barren soil, water-bodies and
vegetation
NDBI = (TM5-TM4) / (TM5+TM4)
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
13
MLC
A pixel-based approach
Attempts to find clusters in the Ndimensional spectral space defined by N
bands of spectral data
ENVI/IDL was used for the test.
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
14
Classification scheme
 build-up area: mixture of urban areas, settlement
or lands under construction;
 cropland: cropland or fallow;
 garden plot: orchards, vineyards or nurseries;
 sparse woodland: low coverage mixture of
shrub, desert scrub or bare ground;
 dense woodland: high coverage mixture of
forest, shrub or shelter belt;
 grassland: pasture or desert grass;
 river flat: dry river bed or river flat;
 water body: reservoirs or fish ponds.
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
15
O-O Image Analysis
Object-oriented approach
Use both spectral and spatial information
eCognition software was used for the test
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
16
O-O classification method
Unclassified ETM+ image
Object oriented image analysis
Image segmentation
Building knowledge base
Referring aerial photos
Training samples selection
Classify image with Nearest
Neighbour (NN) classifier
Classified image in eCognition
2015/7/18
Accuracy assessment
International Conference on China's Urban Land and Housing in the 21 Century
17
Accuracy assessment
Stratified random sampling
Referring to the ortho-corrected aerial
photos.
Totally 900 reference sites are selected as
ground reference points.
Error matrices were created for MLC and
O-O results.
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
18
Results
Classification results from different
approaches
The comparison of accuracy of different
classification
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
19
Result: NDBI
The sparse
woodland, bare
ground and dry
riverbed are
merged into the
same land-cover
class as the
background of
built-up area.
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
20
Result:
MLC
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
21
Result:
O-O
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
22
Error matrix - MLC
Reference Data
Classified Data
On map
On ground
Correct
PA
UA
Conditional
Kappa
1
2
3
4
5
6
7
8
Water body
57
2
1
0
0
0
0
0
60
91
57
62.64%
95.00%
0.9444
Bottomland
8
95
3
1
0
0
0
0
107
103
95
92.23%
88.79%
0.8734
Build-up area
4
5
85
11
12
3
4
0
124
117
85
72.65%
68.55%
0.6385
Sparse woodland
2
0
5
92
14
4
23
1
141
109
92
84.40%
65.25%
0.6046
Cropland
0
0
2
1
126
3
0
39
171
189
126
66.67%
73.68%
0.6669
Garden
0
0
0
0
0
49
0
0
49
90
49
54.44%
100.00%
1.0000
Grassland
0
0
1
4
0
2
59
0
66
86
59
68.60%
89.39%
0.8827
Dense woodland
20
1
20
0
37
29
0
75
182
115
75
65.22%
41.21%
0.3260
Column Total
91
103
117
109
189
90
86
115
900
900
638
Overall Classification Accuracy = 70.89%
Overall Kappa Statistics = 0.6633
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
23
Error matrix – O-O
Reference Data
Classified Data
On map
On ground
Correct
PA
UA
Conditional
Kappa
1
2
3
4
5
6
7
8
Water body
95
4
3
0
1
1
1
0
105
99
95
95.96%
90.48%
0.8930
Bottomland
0
100
0
0
0
0
0
0
100
107
100
93.46%
100.00%
1.0000
Build-up area
2
0
89
0
3
1
17
5
117
105
89
84.76%
76.07%
0.7291
Sparse woodland
0
1
3
104
1
0
0
0
109
107
104
97.20%
95.41%
0.9479
Cropland
0
0
2
1
0
1
152
2
158
198
152
76.77%
96.20%
0.9513
Garden
0
2
1
0
2
0
12
85
102
95
85
89.47%
83.33%
0.8137
Grassland
0
0
7
2
0
88
3
0
100
91
88
96.70%
88.00%
0.8665
Dense woodland
2
0
0
0
91
0
13
3
109
98
91
92.86%
83.49%
0.8147
Column Total
99
107
105
107
98
91
198
95
900
900
804
Overall Classification Accuracy = 89.33%
Overall Kappa Statistics = 0.8773
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
24
Discussion: Comparison between MLC &
O-O
Overall accuracy
Kappa
PA of built-up area
UA of built-up area
2015/7/18
MLC
70.9%
0.663
72.7%
68.6%
International Conference on China's Urban Land and Housing in the 21 Century
O-O
89.3%
0.878
84.8%
76.1%
25
Discussion: usability of NDBI
 The NDBI method is found to be unable to
differentiate urban areas from the background
features such as sparse woodland, bare ground
and dry riverbed in arid regions.
 The usability of such a pixel-based spectral
classifier is severely limited in the arid regions
mainly due to the common presence of landcovers of bare ground and dry riverbed, which
have similar spectral response with built-up
areas.
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
26
Discussion: MLC versus O-O
The object-oriented classifier yields
significantly better overall accuracy than
the MLC method.
For built-up areas, however, the difference
between MLC and O-O methods is less
significant.
Both methods appears to have less
omission errors but larger commission
errors.
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
27
Discussion: shortfalls of the O-O method
 The classification accuracy depends on the
quality of image segmentation. If objects are
extracted inaccurately, subsequent classification
accuracy will not improve.
 Classification error could be accumulated due to
the error in both image segmentation and
classification process.
 Once an object is misclassified, all pixels in this
object will be misclassified.
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
28
Conclusion
 This study has compared classifiers regarding to
built-up area delineation in aridzone of China.
 Although the overall accuracy of the O-O
approach is significantly better than that of MLC,
there is less significant difference for the built-up
area class.
 Further research will be focused on the impact of
spatial resolution of images and the efficiency of
different classifiers.
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
29
2015/7/18
International Conference on China's Urban Land and Housing in the 21 Century
30