See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/264276217

Satellite Image Processing for Land Use and Land Cover Mapping

Article in International Journal of Image, Graphics and Signal Processing · September 2014
DOI: 10.5815/ijigsp.2014.10.03

CITATIONS READS

23 9,527

3 authors, including:

Omkar S N J. Senthilnath
Indian Institute of Science Bangalore Agency for Science, Technology and Research (A*STAR)
164 PUBLICATIONS 4,109 CITATIONS 177 PUBLICATIONS 4,464 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Ashoka Vanjare on 05 July 2017.

The user has requested enhancement of the downloaded file.

I.J. Image, Graphics and Signal Processing, 2014, 10, 18-28
Published Online September 2014 in MECS (http://www.mecs-press.org/)
DOI: 10.5815/ijigsp.2014.10.03

Satellite Image Processing for Land Use and

Land Cover Mapping
Ashoka Vanjare1+, S.N. Omkar1*, J.Senthilnath1
1
Department of Aerospace Engineering, Indian Institute of Science, Bangalore
Corresponding Authors Email: [email protected], *[email protected]

Abstract—In this paper, urban growth of Bangalore detection are fuel type assessment [6], and urban growth
region is analyzed and discussed by using multi-temporal analysis [7-9] were studied.
and multi-spectral Landsat satellite images. Urban In this paper, urban growth analysis of Bangalore
growth analysis helps in understanding the change region is analyzed by using multi-temporal satellite
detection of Bangalore region. The change detection is images. Increasing population has resulted in rapid
studied over a period of 39 years and the region of urbanization. This has led to an increase in the need for
interest covers an area of 2182 km2. The main cause for infrastructure development which in turn has resulted in
urban growth is the increase in population. In India, rapid traffic congestion and basic amenities [2]. Hence demand
urbanization is witnessed due to an increase in the for infrastructure has resulted in exploitation of natural-
population, continuous development has affected the resources. The continuous exploitation of natural
existence of natural resources. Therefore observing and resources has led the researchers to study significant
monitoring the natural resources (land use) plays an change detection caused by human impact.
important role. To analyze changed detection, The monitoring of land use and land cover (LULC) is
researcher’s use remote sensing data. Continuous use of carried-out by using remote sensing data which involves
remote sensing data helps researchers to analyze the the use of several multi-date and multi-sensor satellite
change detection. The main objective of this study is to images [10-12]. The successful uses of remote sensing
monitor land cover changes of Bangalore district which data are obtained by having an adequate understanding of
covers rural and urban regions using multi-temporal and landscape features [13] and imaging systems [14].
multi-sensor Landsat - multi-spectral scanner (MSS), Satellite data are the most common data sources for
thematic mapper (TM), Enhanced Thematic mapper plus studying change detection [15] and region mapping [13].
(ETM+) MSS, TM and ETM+ images captured in the It is used because of its repetitive data acquisition, data
years 1973, 1992, 1999, 2002, 2005, 2008 and 2011. storage, digital processing [14] and analyzed the content.
Temporal changes were determined by using maximum In literature, change detection techniques [16] and
likelihood classification method. The classification classification accuracy methods [13-16] have been
results contain four land cover classes namely, built-up, devised by researchers to study LULC mapping problems.
vegetation, water and barren land. The results indicate Urban settlements are common in all parts of the world
that the region is densely developed which has resulted in and it is growing at a very high pace. In India, cities such
decrease of water and vegetation regions. The continuous as Delhi, Mumbai, Calcutta, Chennai and Bangalore are
transformation of barren land to built-up region has witnessing a significant urban growth due to improper
affected water and vegetation regions. Generally, from urban planning. In these urban regions, Bangalore region
1973 to 2011 the percentage of urban region has is widely affected due to rapid industrialization. The city
increased from 4.6% to 25.43%, mainly due to is witnessing exploitation of natural resources [17] in
urbanization. terms of decrease in number of wetlands and also
decrease in rainwater catchment areas. So these factors
Index Terms—Urban growth analysis, Land use, Land have lead researchers to use multi-temporal satellite
cover, Change detection, Multi-temporal satellite images. images in order to study the significant changes that have
occurred due to human activities.
Researchers have used supervised and unsupervised
I. INTRODUCTION techniques [18-21] to classify satellite images [9-12].
Image classification is a technique to categorize all the
Multi-temporal satellite images provide an excellent
image pixels into group of similar classes. Remote
spatial temporal features [1] and it is also freely available
sensing data are used to perform classification which is
in the public domains [2]. Multi-temporal satellite images based on the spectral patterns. The objective of image
[3] are widely used in applications for studying both long classification [22-29] is to identify the unique features
term and short term change detection. Some of the
occurring in an image that actually represent on the
applications in change detection – in case of short term
ground features. Bakr et al. [1] used multi-temporal
change detection are crop yield estimation [4] and flood
Landsat data for observing land cover changes. In their
assessment [5], and in case of long term temporal change
study, they have used maximum likelihood and Iterative

Copyright © 2014 MECS I.J. Image, Graphics and Signal Processing, 2014, 10, 18-28
 Satellite Image Processing for Land Use and Land Cover Mapping 19

self-organizing maps methods to classify satellite images. improve the image quality. The main steps in pre-
Dengsheng Lu et al. [12] successfully used Landsat processing [1-12] involve resampling, geo-referencing,
images for studying impervious surface change detection sub-setting and cloud removal [28-35]. All satellite
in urban-rural frontier using supervised linear spectral images are geo-referenced using known ground control
mixture analysis method. Anderson et al. [10] has points (GCP’s) and the given datasets are projected to
proposed different levels (single and multi-level) of world geodetic system (WGS) - 84 (Zone-UTM 43 North)
classification system for LULC mapping problems. as the datum. Landsat ETM+ experienced a failure of its
Dengsheng et al. [12] successfully used linear spectral Scan Line Correcting (SLC) on May, 2003 and it showed
mixture analysis in order to detect impervious surface in permanent disability in SLC. In LANDSAT ETM+
urban-rural frontiers. Javed Mallick et al. [21] estimated images, gap filling [31-32] is carried out by copying
land surface temperature from Landsat-7 ETM+ images adjacent pixels for correcting SLC method. Filling the
using maximum likelihood classification and minimum scan gap requires precise knowledge, so adjacent pixels
noise fraction methods for classifying satellite images. are copied into missing pixels. The resultant of gap filling
El-Kawya et al. [22] made use of image enhancement and is shown in the Fig.1.
visual interpretation techniques in order to improve
supervised classification on Landsat images. Camps-
Valls et al. [23] designed a framework for supervised
method using support vector machine based on kernels
for multi-temporal and multi-source remote sensing data
classification. Esch T et al. [24] have used satellite
images for determining impervious surface using support
vector machines algorithm. Weber C et al. [25] have used
a stepwise discriminate analysis, a supervised method for
land cover classification. Avic et al. [26] presented Fig 1: (a) Satellite image of Landsat 7 showing before scan line
corrector failure and (b) Satellite image of Landsat 7 showing after scan
hierarchical classification method to classify Landsat TM line corrector failure.
imagery in order to map land cover regions.
Ramachandra T.V et al. [2] have used Landsat images In this paper, three different sensors, namely, multi-
of greater Bangalore district with an approximate area spectral scanner (MSS), thematic mapper (TM),
[27] of 61 km2 for studying urban growth. In their study, Enhanced Thematic mapper plus (ETM+) are used for
they have used bayesian networks to classify the images studying change detection for Bangalore region. Here,
and observed significant decrease in the number of water except thermal band we have used all the multi-spectral
bodies. In our study, we are considering an area of 2182 bands of 1973, 1992, 1999, 2002, 2005, 2008, 2010 and
Km2 for analysis of Bangalore urban growth. Here, we 2011 for classification. The different satellite sensors are
have used different satellite sensors like LANDSAT MSS, used in this paper are given below:
TM, and ETM+ sensor images for studying Bangalore
urban growth. The multi-temporal satellite images with a 1) Landsat MSS with 4 bands and 79 mtrs resolution
period of 39 years are used for this study. acquired on 27th Feb 1973.
Initially, the images are pre-processed [30-59] using 2) Landsat TM with 7 bands and 30 mtrs resolution
different image processing techniques for analysis of land acquired on 14th Jan 1992.
cover regions. These images are classified using 3) Landsat TM with 7 bands and 30 mtrs resolution
maximum likelihood classification method to produce acquired on 7th April 1999.
thematic map into four classes - built-up, vegetation, 4) Landsat ETM+ with 8 bands and 30 mtrs resolution
water and barren lands. Thematic classification results acquired on 18th Feb 2002.
assist researchers in determining the number of pixels of 5) Landsat ETM+ with 8 bands and 30 mtrs resolution
each region for studying the change detection. The main acquired on 25th Jan 2005.
objectives of this study are: 1) to provide a recent 6) Landsat ETM+ with 8 bands and 30 mtrs resolution
perspective for different land cover types of Bangalore acquired on 18th Jan 2008.
region, and 2) monitoring land cover changes of 7) Landsat ETM+ with 8 bands and 30 mtrs resolution
Bangalore region from the year 1973 to 2011 by using acquired on 23rd Jan 2010.
maximum likelihood classification technique. 8) Landsat ETM+ with 8 bands and 30 mtrs resolution
This paper is divided into following sections, section 2 acquired on 31st March 2011.
details data preparation technique, image classification
techniques are given in section 3, results and discussions A. Resampling
are given in section 4 and section 5 presents the
conclusions. Initially all the given satellite images are re-sampled
[33]. This is a process in which all the different images of
different resolution are converted into a standard
II. DATA PREPARATION resolution [1]. In re-sampling, the root means square
error (RMSE) is reported to be less than 0.3 pixels.
In this section, the image pre-processing methods are RMSE value between any two different dates and
discussed. Initially, this pre-processing stage is used to different sensor is acceptable if it is less than 0.5 pixels.

Copyright © 2014 MECS I.J. Image, Graphics and Signal Processing, 2014, 10, 18-28
20 Satellite Image Processing for Land Use and Land Cover Mapping

The RMSE error value after re-sampling for the data used where f 'ref is the new value, f ref is the brightness value
in this study is 0.3 pixels which is less than 0.5 pixels. So
this data set can be used for classification. of a pixel of the reference image, mref is the mean of the
B. Geo-correction reference image,  ref is the standard deviation of the
Landsat datasets are geo-corrected with known ground reference image, mmain is the mean of the clouded image
control points (GCP). For geometric correction we have
considered 2011 satellite image as reference images and  main is the standard deviation of the clouded image.
where all other images used in this work are re-sampled Here the clouded image is considered to be an original
and co-registered to the reference image. Geo-correction image which is adjusted using the reference image. So
is done using both topo-maps and ground control points that, it’s mean and standard deviation match the clouded
(GCP). Each satellite images are geo-referenced by image. Clouds reflect [30] the solar radiation in the
keeping 100 GCP (ground control points) for all the visible and infrared spectra to a much higher degree than
images (1973-2011). open grounds so we set a threshold C1 (C1=250) to
distinguish between cloud regions and the open ground
C. Subsetting region of interest regions (where open ground region pixel value is nearly
Survey of India top sheets of 1:50,000 and 1:2, 50,000 150). The value of the threshold C1 is determined by
scales [35-38] were used to generate vector layer (base histogram values of the image. Further, we compute
layer). The vector layer is prepared by digitizing the absolute value in order to determine the reliable detection
administration boundary and it is overlaid on raster layers (i.e. difference between the same locations of the main
(satellite images). Here the geo-referenced data (both image and the brightness-corrected reference image) is
raster image and vector layers) are used to subset the given by,
region of interest. In the Fig. 2, region of interest is
cropped from given satellite image (full swath). f main (i, j)  c1  f 'ref (i, j)  c1 | f main (i, j)  f 'ref (i, j)| c2
(2)

where f 'ref is the reference image, f main is the main

image, mref is the mean of the reference image,  ref is
the standard deviation of the reference image, mmain is
the mean of the main image and  main is the standard
Fig 2: (a) Shape File of Bangalore, (b) Bangalore Shape File overlaid on
the Original Image, (c) Bangalore district is cropped from Original deviation of the main image A threshold C2 ( C2=200 ) is
Image used to ensure the reliability and validation of the
detected region in clouded region. A binary decision map
D. Cloud Removal is generated by using two thresholds values C1 and C2. To
identify cloud shadow pixels in case of satellite bands-
Generally, in multi-spectral imaging, fusion techniques gray format is very hard (due to brightness value of
are used to remove cloud pixels from the satellite images shadow is quite close to other regions and it is much
by replacing each pixels of the main image by using a smoother than the other those of the other surrounding
reference image. For LANDSAT ETM+ 2011 image, the regions). The cloud-free pixels from 2010 image
isolated cloud pixels were found by visual inspection. So information are used to replace 2011 image clouded
we used image fusion method for removing cloud pixels pixels. After all the pre-processing steps the final data
[30-39]. Image fusion method is a process of combining sets are shown in the Fig.3. From the Fig.3, we can make
information from two or more images in order to create out visually all the clouded pixels are removed.
another image [13-15]. The six bands of Landsat ETM+
2010 nearest date image are used for replacing Landsat
ETM+ 2011 image for cloud pixel replacing. For each
band, we have corrected the brightness and a threshold is
used for detecting the cloud pixels. The different solar
irradiance and atmospheric effects are eliminated from
each spectral band by assuming linear relation [29-32]
between the corresponding brightness values of the two
images.
The equation which is used for brightness correction is
given by,
Fig 3: (a) Clouded image of Landsat ETM+ 2011, (b) Cloud free image
of Landsat ETM+ 2010, (c) Clouded image pixels replaced in Landsat
 main ETM+ 2011
f 'ref (i, j )  [ f ref (i, j )  mref ]  mmain (1)
 ref

Copyright © 2014 MECS I.J. Image, Graphics and Signal Processing, 2014, 10, 18-28
 Satellite Image Processing for Land Use and Land Cover Mapping 21

matrix for each class i with the size of m*m and n is the
III. IMAGE CLASSIFICATION METHOD number of classes. MLC [16] is considered as standard
classification algorithm for comparing with the other
Image classification methods [36-40] are very useful in algorithms. The samples are represented by cluster space
identifying different features from the given image.
representation where information (or samples) is
Features like built-up, water, vegetation and barren land
automatically assigned to their spectral classes. Pixel
can be used for exploring in order to understand the
labelling is carried-out by combined decision based on its
multi-temporal variations. Multi-temporal satellite
relationship to the defined clusters and cluster
images [6-12] provide excellent temporal variations membership of the belonging information classes. The
which can be used for urban growth analysis. Different accuracy assessment is carried-out using known ground
combinations of bands [13-15] are generated in order to
control points (GCP’s) and for cross verifying we have
identify built-up, vegetation, water and barren land
used Google Earth® tool. Google Earth data
signatures from the satellite images (signatures means
(http://earth.google.com) is used for pre-classification
similar spectral values). The supervised classification and post-classification processes and validation of the
methods are used for pattern classification [16]. results. The results are compared and analyzed using
Supervised classification [16] identifies class information
classification matrix.
in the satellite images and similar pixels are used as
‘training samples’ (signature values). The classifier
system is used to determine the statistical
IV. RESULTS AND DISCUSSIONS
characterization of reflectance for each information class
and this stage is called ‘Signature analyses’. Signature In this section, descriptions of the study area and land
analyses involve statistical characterization of the range cover change analysis are discussed. The Study area-
of reflectance on each band. The statistical Bangalore region is located in the south of Karnataka
characterization has been achieved for each information state. It is the capital city of Karnataka state. It is India’s
class. Then the image is classified by examining the third most populous city and also it is fifth most populous
reflectance for each pixel and making a decision about urban agglomeration. The latitude and longitude of
which of the signature it resembles accurately [13-18]. central Bangalore region is 12°58′0″N and 77°34′0″E.
The band combinations for each image are shown
below which are used to collect signatures or training
samples from the given datasets:

1) Landsat-MSS 1973 data with band combinations

(for false color composite (FCC) - 321 band and
True color composite (TCC) - 432 band)
2) Landsat-TM 1992, 1999 data with band
combinations (for FCC - 432 band and TCC 321
band)
3) Landsat-ETM+ 2002, 2005, 2008, 2011 data with
band combinations (for FCC 432 band and TCC 321
band) Fig 4: Study of Bangalore urban region is shown.

In Fig.4, the band combinations are used for collecting

Bangalore receives moderate rainfall [17]. The mean
training samples. The collected samples are given as
annual total rainfall is about 880 mm with 60 rainy days a
input to maximum likelihood classification method.
year over a period of ten years from 2000 to 2010. The
Satellite image are classified using ERDAS® - classifier
temperature in summer is about 18º- 38ºand while winter
module. Maximum Likelihood Classifier (MLC) is a
is about 12º- 25º. Bangalore is located at an attitude of
well-known parametric statistical classifier algorithm
920 mtrs above the mean sea level. This has created an
[16]. In MLC, normal distribution is assumed for the
undulating terrain in the Bangalore region and has
input data. The input data consists of two parameters – a
facilitated in creation of large number of wetlands. Fig.4
mean vectors and a covariance matrix for estimation of
shows study area of Bangalore region.
class distributions. The mean and covariance matrix is
Maximum likelihood classification is used for
used in discriminant functions to find discrimination in
classifying the given satellite images which results in
the distribution. The discriminant function in MLC is
thematic images. The thematic images are used to
given by:
monitor and detect land cover changes in the Bangalore
urban region, from the given different satellite images for
1 (3)
g ( x)   ln   ( X  m )T  ( X  m ) the years 1973 to 2011. Maximum likelihood
i i i classification [16-18] method is applied for all the
i i
satellite datasets for digital data exploration. The satellite
where X is the m dimensional input feature vector, mi is images are classified to produce four before mentioned
the mean feature vector for class i, Σi is the covariance thematic classes. The thematic classified images are used

Copyright © 2014 MECS I.J. Image, Graphics and Signal Processing, 2014, 10, 18-28
22 Satellite Image Processing for Land Use and Land Cover Mapping

to analyze the changes that occurred from the year 1973 Table 2: Converted the Landsat satellite image pixels into km2 where
we can see vegetation and water is decreasing in terms of kilometer.
to 2011. The classification method is applied for data
analysis. Table 1 explains the thematic information SL Built-up Vegetation Water Barren
obtained from the classification technique. Table 2 YEAR
NO Km2 Km2 Km2 Km2
explains the area coverage for each land cover class by
square kilometer percentage across several dates. The 1 1973 101.7918 386.3457 88.6248 1604.7
resulting thematic land cover change images are shown in
Fig.5 - 11. Further, the classification matrix is used to
2 1992 304.3755 322.6941 76.6791 1478.11
analyze the urban growth. The classified images are
interpreted in order to obtain number of pixels for
calculating area. Temporal analysis helps in studying the 3 1999 383.931 174.1221 27.3267 1596.08
change detection for analyzing decrease in number of
water and vegetation image pixels. The results shows that 4 2002 510.2019 152.6562 24.4251 1493.61
in the years 1973, 1992, and 1999 land cover is slowly
increasing in Bangalore district with a total area coverage
5 2005 527.1588 151.0623 9.5976 1493.98
from 101.80 km2 to 383.93 km2. But a sudden increase in
the urban region can be seen during the years 2002 to
2011 from 510.20 km2 to 554.82 km2. 6 2008 535.6584 150.9768 9.5805 1485.58
From table 2, it can be seen that during the year 1973,
area covered by water is 88.62 km2 and area covered by 7 2011 554.8275 78.3126 9.6111 1539.04
vegetation is 386.34 km2 while in the year 2011, water is
9.6 km2 and vegetation is 78.31 km2. As a result of urban For the year 1992, area covered by vegetation land is
region development in this region, a drastic increase in 14.8% (322.70 km2) and area covered by water region is
urban area is observed between 1973 to 2011. In Fig. 5- 3.48% (75.9 km2) for the Bangalore region. Urban region
11, different satellite images and classified images are has also increased in comparison with the year 1973; it
shown. In Fig. 5, it can be seen that built-up area is about has covered 13.95% (304.37 km2). Significant increase in
101.1 km2 and in Fig. 11 it is observed that built-up has urban regions can be noticed for the year 2002, 2005 and
increased to 1554.8 km2. From table 2, it is seen that 2008 and also sudden decrease in water and vegetation
water region is decreasing during the year 2002 to 2011 regions. In the year 2011, similar trends were observed.
and vegetation region is also following the same pattern. Vegetation land has decreased to 3.6% (78.31 km2),
In the year 2002, area covered by vegetation region is water region is 0.6% ( 9.61 km2) and urban region is
152.65 km2 and where as 2011, area covered by 25.40% (554.8 km2). Here it can be observed from the
vegetation region is 78.31 km2. image, the interpreted results of the vegetation and water
regions have decreased drastically. From the year 1973
Table 1: Landsat satellite image classification where four different image, a large number of water bodies can be seen and it
classes built-up, vegetation, water and barren.
has gradually decreased in the year 1992, 1999, 2002,
2005, 2008 and 2011. In Fig.12-15, the four important
classes, namely, built-up, water, vegetation and barren
SL Built-up Vegetation Water Barren
NO
YEAR
Pixels Pixels Pixels Pixels
land are shown using bar graph. From the graph, we can
infer that built-up region has increased which has resulted
in a decrease of vegetation and water regions. The results
shows that there is sudden increase of land cover change
1 1973 113102 429273 98472 1782995
rate between 1973 and 2002. Further it has increased
more in 2011 image. The unplanned urbanization has
2 1992 338195 358549 85199 1642342 affected the existence of water and vegetation regions
further the natural catchments have been affected due to
3 1999 426590 193469 30363 1773425 urban growth. This will affect water logging in these
regions. From the results, it can be infered that the water
regions have disappeared over a period of time due to
4 2002 566891 169618 27139 1659567
the human impact.

5 2005 585732 167847 10664 1659974

V. CONCLUSIONS
6 2008 595176 167752 10645 1650644 This paper presents the result of Bangalore urban
region for over 39 years (1973-2011). In this research,
7 2011 616475 87014 10679 1710049 maximum likelihood pattern classification algorithm is

Copyright © 2014 MECS I.J. Image, Graphics and Signal Processing, 2014, 10, 18-28
 Satellite Image Processing for Land Use and Land Cover Mapping 23

used in order to analyze temporal satellite images. The ACKNOWLEDGEMENT

results clearly indicated the decrease in number of
This work is supported by the Space Technology Cell,
vegetation image pixels and water image pixels with the
Indian Institute of Science and Indian Space Research
increase in number of urban pixels. The year 1973 had a
Organization grant. Satellite data for experimentation are
large number of water regions in comparison with 1992
distributed by the Land Processes Distributed Active
and 2011. As urban regions is growing and it has affected
Archive Center (LP DAAC), located at the U.S.
the natural resources like water and vegetation. The
Geological Survey (USGS) Earth Resources Observation
influence and impact of human settlements on the
and Science (EROS) Center (lpdaac.usgs.gov).
environment can be seen and justified.

Legends
Built-up Land
Barren Land
Water Land
Vegetation Land

Fig 5: a) TCC b) FCC and c) Classified Landsat MSS 1973 satellite image.

Fig 6: a) TCC b) FCC and c) Classified Landsat TM 1992 satellite image.

Fig 7: a) TCC b) FCC and c) Classified Landsat 7TM 1999 satellite image.

Copyright © 2014 MECS I.J. Image, Graphics and Signal Processing, 2014, 10, 18-28
24 Satellite Image Processing for Land Use and Land Cover Mapping

Fig 8: a) TCC b) FCC and c) Classified Landsat ETM+ 2002 satellite image.

Fig 9: a) TCC b) FCC and c) Classified Landsat ETM+ 2005 satellite image.

Fig 10: a) TCC b) FCC and c) Classified Landsat ETM+ 2008 satellite image.

Fig 11: a) TCC b) FCC and c) Classified Landsat ETM+ 2011 satellite image.

Copyright © 2014 MECS I.J. Image, Graphics and Signal Processing, 2014, 10, 18-28
 Satellite Image Processing for Land Use and Land Cover Mapping 25

multi-temporal Landsat data, Applied Geography,

30(4), 592-605.Elsevier Ltd. Doi:
10.1016/j.apgeog.2009.10.008.
[2] Ramachandra T. V and Uttam Kumar Wetlands of
Greater Bangalore, India: Automatic Delineation
through pattern classifiers, Electronic Green Journal,
(26), pp-1-22.
[3] Walkey, J. A. Development of a change detection
tool for image analysis, MS thesis. University of
Wisconsin-Madison, 1997.
[4] Prasad, A. K., Chai, L., Singh, R. P., and Kafatos,
Fig 12: Landsat satellite image classifications results for different year
M. Crop yield estimation model for Iowa using
is shown in bar charts for built up region. remote sensing and surface parameters,
International Journal of Applied Earth Observation
and Geoinformation, 8, 26-33. Doi:
10.1016/j.jag.2005.06.002.
[5] Bhavsar, P. D. Review of remote sensing
applications in hydrology and water resources
management in India, Advances in Space Research.
4 (11), 193-200, 1984.
[6] Van, W. J. W., Root and R. R. Hyper spectral
analysis of multi-temporal Landsat TM data for
mapping fuels in Yosemite National Park, Proc.
Joint Fire Sci. Conf. and Workshop. Boise, Idaho.
June 1999
Fig 13: Landsat satellite image classifications results for different year
[7] Zhang, X., Sun, R., Zhang, B. and Tong, Q. Land
is shown in bar charts for Vegetation region. cover classification of the North China Plain using
MODIS EVI time series, ISPRS Journal of
Photogrammetry and Remote Sensing. 63, 476–484,
2008.
[8] Verhoest, N., Troch, P. A., De T. F.P. Spatial soil
moisture mapping through multi-temporal analysis
of ERS-SAR PRI data, 3rd ERS SYMPOSIUM,
Florence, 1997.
[9] Ashbindu Singha, Digital change detection
techniques using remotely-sensed data, International
Journal of Remote Sensing, Volume-10(6), pp 989-
1003, 1989.
Fig 14: Landsat satellite image classifications results for different years [10] James r. Anderson, Ernest e. hardy, John t. roach,
is shown in bar charts for water region. and Richard e. witmer, a land use and land cover
classification system for use with remote sensor data,
1972.
[11] El-Kawya, O. R. A., Rod, J. K., Ismail, H. A., and
Suliman A. S, Land use and land cover change
detection in the western Nile delta of Egypt using
remote sensing data, Applied Geography, 31(2),
483-494. Elsevier Ltd. Doi:
10.1016/j.apgeog.2010.10.012.
[12] Dengsheng Lu, Emilio Moran and Scott Hetrick,
Detection of impervious surface change with
Multitemporal Landsat images in an urban–rural
frontier, ISPRS Journal of Photogrammetry and
Fig 15: Landsat satellite image classifications results for different years Remote Sensing, 26 Nov 2010.
is shown in bar charts for barren land region.
[13] Robert A. Schowengerdt, Remote Sensing: Models
and Methods for Image Processing, third edition.
[14] Gonzalez, R. C. and Woods, R. E, Digital Image
REFERENCES
Processing, 3E, Prentice Hall, Upper Saddle River,
[1] Bakr N, Weindorf, D C, Bahnassy, M. H, Marei, S NJ.
M., & El-badawi M. M. Monitoring land cover [15] Thomas Lille sand, Ralph W. Kiefer and Jonathan
changes in a newly reclaimed area of Egypt using Chipman, Remote Sensing and Image Interpretation.

Copyright © 2014 MECS I.J. Image, Graphics and Signal Processing, 2014, 10, 18-28
26 Satellite Image Processing for Land Use and Land Cover Mapping

[16] R. O. Duda,P. E. Hart and D. G. Stork, Pattern Computation, 205(2), 584-600. Elsevier Inc. doi:
Classification 2E John Wiley & Sons, 2000. 10.1016/j.amc.2008.05.050.
[17] Ramachandra T. V, & Kumar Uttam, Greater [31] M.J. Pringle,M. Schmidta and J.S. Muira,
Bangalore: Emerging Urban Heat Island, Urban Geostatistical interpolation of SLC-off Landsat
Ecosystems. ETM+ images , ISPRS Journal of Photogrammetry
[18] Ethem Alpaydm, Introduction to Machine Learning, and Remote Sensing ,Volume 64(6), pp 654-
MIT Press 664,2009.
[19] Lu, D., & Weng, Q. (2006), Use of impervious [32] Gyanesh Chander, Brian L. Markham and Dennis L.
surface in urban land-use classification, Remote Helder, Summary of current radiometric calibration
Sensing of Environment, 102, 146 - 160. Doi: coefficients for Landsat MSS, TM, ETM+, and EO-
10.1016/j.rse.2006.02.010. 1 ALI sensors , Remote Sensing of Environment,
[20] Esch, T., Himmler, V., Schorcht, G., Thiel, M., Volume 113(5), pp 893-903.
Wehrmann, T., Bachofer, F., Large-area assessment [33] Song, C., Woodcock, C. E., Seto, K. C., Lenney, M.
of impervious surface based on integrated analysis P., and Macomber, S. A, Classification and Change
of single-date Landsat-7 images and geospatial Detection Using Landsat TM Data: When and How
vector data, Remote Sensing of Environment, 113, to Correct Atmospheric Effects? Data Processing,
1678-1690. Doi: 10.1016/j.rse.2009.03.012. 4257(00).
[21] Mallick, J., Kant, Y., and Bharath, B. D, Estimation [34] Kang-tsung Chang, Introduction to Geographic
of land surface temperature over Delhi using Information Systems.
Landsat-7 ETM +, 12(3), 131-140. [35] S. N. Omkar, J. Senthilnath, Dheevatsa Mudigere
[22] El-Kawya, O. R. A., Rod, J. K., Ismail, H. A., & and M. Manoj Kumar, Crop Classification using
Suliman, A. S, Land use and land cover change Biologically-inspired Techniques with High
detection in the western Nile delta of Egypt using Resolution Satellite Image, Journal of the Indian
remote sensing data, Applied Geography, 31(2), Society of Remote Sensing Volume 36(2), 175-182.
483-494. Elsevier Ltd. doi: [36] Adel Shalaby and Ryutaro Tateishi, Remote sensing
10.1016/j.apgeog.2010.10.012. and GIS for mapping and monitoring land cover and
[23] Camps-valls, G., Gómez-chova, L., Munoz-Mari, J., land-use changes in the Northwestern coastal zone
Rojo-álvarez, J. L., Martinez-Ramón, M., (2008), of Egypt, Applied Geography, Volume 27(1), and
Kernel-Based Framework for Multi-temporal and Pages 28-41, 2007.
Multi-source Remote Sensing Data Classification [37] Peter Potapov, Svetlana Turubanova, and Matthew
and Change Detection, 46(6), 1822-1835. C. Hansen, Regional- Scale boreal forest cover and
[24] Esch, T., Himmler, V., Schorcht, G., Thiel, M., change mapping using Landsat data composites for
Wehrmann, T., Bachofer, F, Large-area assessment European Russia, Remote Sensing of Environment,
of impervious surface based on integrated analysis October 2010.
of single-date Landsat-7 images and geospatial [38] Hugo Carrão, Paulo Gonçalves and Mário Caetano,
vector data, Remote Sensing of Environment, 113, Contribution of multispectral and Multitemporal
1678-1690. Doi: 10.1016/j.rse.2009.03.012. information from MODIS images to land cover
[25] Weber, C., and Puissant, A, Urbanization pressure classification, Remote Sensing of Environment,
and modeling of urban growth: example of the Volume 112(3), PP 986-997, 2008.
Tunis Metropolitan Area. Remote Sensing of [39] David P. Roy, Junchang Ju, Philip Lewis, Crystal
Environment, 86, 341 - 352. Doi: 10.1016/S0034- Schaaf, Feng Gao, Matt Hansen and Erik Lindquist,
4257(03)00077-4. Multi-temporal MODIS–Landsat data fusion for
[26] Lu, D., Batistella, M., and Moran, E, Multitemporal relative radiometric normalization, gap filling, and
spectral mixture analysis for Amazonian land-cover prediction of Landsat data , Remote Sensing of
change detection Methods, 30(1), pp 87-100. Environment, Volume 112, Issue 6(16) Pages 3112-
[27] Avci, M., & Akyurek, Z, A Hierarchical 3130,2008.
Classification of Landsat-TM Imagery for Land [40] J. Senthilnath, S.N. Omkar, V. Mani and T.
covers Mapping. Karthikeyan, Multi-objective Discrete Particle
[28] Ramachandra T. V. and Uttam Kumar, Greater Swam Optimization for Multi-sensor Image
Bangalore: Emerging Urban Heat Island, Energy Alignment" IEEE Geoscience and Remote Sensing
and Wetlands Research Group Centre for Ecological Letters (GRSL), Vol. 10(5), pp. 1095 - 1099, 2013.
Sciences, Indian Institute of Science, Bangalore – [41] J.Senthilnath, S.N. Omkar, V. Mani, Naveen P
560 012, INDIA, http://www.gisdevelopment.net. Kalro and P.G. Diwakar, Multi-objective genetic
[29] Wang, B., Ono, A., and Muramatsu, K, Automated algorithm for efficient point matching in multi-
Detection and Removal of Clouds and their sensor satellite image” Proc. IEEE International
Processing”, (2), 453-460. Geoscience and Remote Sensing Symposium
[30] Tseng, D.-chang, Tseng, H.-ting and Chien, C- (IGARSS'12), Munich, Germany, 2012.
Liang, Automatic cloud removal from multi- [42] J.Senthilnath, S.N. Omk ar, V. Mani and T.
temporal SPOT images, Applied Mathematics and Karthikeyan, Multi-objective optimization of

Copyright © 2014 MECS I.J. Image, Graphics and Signal Processing, 2014, 10, 18-28
 Satellite Image Processing for Land Use and Land Cover Mapping 27

satellite image registration using discrete particle [53] J. Senthilnath, Ankur raj, S.N. Omkar, V. Mani and
swarm optimisation Proc. IEEE INDICON'11, Deepak kumar, Quasi-Based Hierarchical Clustering
Hyderabad, India, 2011. for Land Cover Mapping Using Satellite Images,
[43] J. Senthilnath, Shivesh Bajpai, S.N. Omkar, P.G. Proceedings of Seventh International Conference on
Diwakar and V. Mani An approach to Multi- Bio-Inspired Computing: Theories and Application
temporal MODIS Image analysis using Image Advances in Intelligent Systems and Computing,
classification and segmentation, Advances in Space Springer India, Vol. 202, pp. 53 – 64, 2013.
Research, Vol. 50(9), pp. 1274 - 1287, 2012. [54] J. Senthilnath, Vikram Shenoy H, S.N. Omkar and
[44] C.S. Arvind, Ashoka Vanjare, S.N. Omkar, J. V. Mani, Spectral-spatial MODIS image analysis
Senthilnath, V. Mani and P.G. Diwakar, Multi- using Swarm intelligence algorithms and region
temporal Satellite Image Analysis Using based segmentation for flood assessment,
Unsupervised Techniques, Advances in Computing Proceedings of Seventh International Conference on
and Information Technology, (Eds. N. Meghanathan Bio-Inspired Computing: Theories and Application
et al.) Advances in Intelligent Systems and (Eds. J.C. Bansal et al.) Advances in Intelligent
Computing, Springer Verlag, Berlin, Germany, vol. Systems and Computing, Springer India, Vol. 202,
177, pp. 757 – 765, 2013. pp. 163 – 174, 2013.
[45] J. Senthilnath, Vikram Shenoy H, Ritwik Rajendra, [55] J. Senthilnath, Vipul Das, S.N. Omkar and V. Mani,
S.N. Omkar, V. Mani and P.G. Diwakar, Integration Clustering using Levy Flight Cuckoo Search
of speckle de-noising and image segmentation using Proceedings of Seventh International Conference on
Synthetic Aperture Radar image for flood extent Bio-Inspired Computing: Theories and Application
extraction, Journal of Earth System Science, (Eds. J.C. Bansal et al.) Advances in Intelligent
(Springer-Verlag), Vol. 122(3), pp. 559 - 572, 2013. Systems and Computing, Springer India, Vol. 202,
[46] J. Senthilnath, S.N. Omkar, V. Mani, Nitin Karnwal pp. 65 – 75, 2013.
and Shreyas P.B, Crop Stage Classification of [56] J. Senthilnath, Shreyas P.B, Ritwik Rajendra, S.N.
Hyperspectral Data using Unsupervised Techniques Omkar, V. Mani and P.G. Diwakar, Multi-Sensor
IEEE Journal of Selected Topics in Applied Earth Satellite Image Analysis using Niche Genetic
Observations and Remote Sensing (IJSTARS), Vol. Algorithm for Flood Assessment, Swarm,
6(2), pp. 861 - 866, 2013. Evolutionary and Memtic Computing Conference,
[47] J. Senthilnath, S.N. Omkar, V. Mani and P.G. Springer-Verlag Berlin Heidelberg, LNCS 7677, pp.
Diwakar, Multi-temporal Satellite Imagery for 49 - 56, Springer, Heidelberg, 2012.
Flood Damage Assessment, Journal of the Indian [57] J. Senthilnath, S.N. Omkar, V. Mani and Nitin
Institute of Science, Special Issue on imaging and Karnwal, Hierarchical artificial immune system for
microscopy, Vol. 93(1), pp. 105 - 116, 2013. crop stage classification, Proc. IEEE INDICON'11,
[48] J. Senthilnath, S.N. Omkar, V. Mani, Tejovanth N, Hyderabad, India, 2011.
P.G. Diwakar and Archana Shenoy B, Hierarchical [58] J. Senthilnath, S.N. Omkar, V. Mani, Tejovanth N,
Clustering Algorithm for Land Cover Mapping P.G. Diwakar and Archana Shenoy B, Multi-
using Satellite Images, IEEE Journal of Selected Spectral Satellite Image Classification using
Topics in Applied Earth Observations and Remote Glowworm Swarm Optimization, Proc. IEEE
Sensing, Vol. 5(3), pp. 762 - 768, 2012. International Geoscience and Remote Sensing
[49] J. Senthilnath, S.N. Omkar and V. Mani, Clustering Symposium (IGARSS'11), Vancouver, Canada,
using firefly algorithm: performance study Swarm 2011.
and Evolutionary Computation, Vol. 1(3), pp.164 - [59] M. Rajeswari, K.S. Gurumurthy, S.N. Omkar, J.
171, 2011. Senthilnath and L. Pratap Reddy, Automatic Road
[50] S.N. Omkar, Sivaranjani V, J. Senthilnath and Extraction using High Resolution Satellite Images
Suman Mukherjee, Dimensionality Reduction and based on Level Set and Mean Shift Methods,
Classification of Hyperspectral Data, International Proc. IEEE 3rd International Conference on
Journal of Aerospace Innovations, Vol. 2(3), pp. Electronics Computer Technology (ICECT'11),
157 - 163, 2010. Kanyakumari, India, 2011.
[51] J. Senthilnath, M. Rajeswari and S.N. Omkar,
Automatic Road Extraction using High Resolution
Satellite Image based on Texture Progressive
Analysis and Normalized Cut method, Journal of AUTHORS
the Indian Society of Remote Sensing, (Springer-
Verlag), Vol. 37(3), pp. 351 - 361, 2009. Mr. Ashoka Vanjare, male is a
[52] S.N. Omkar, J. Senthilnath, Dheevatsa Mudigere project assistant at Indian institute of
and Manoj Kumar M, Crop Classification using science, Bangalore. He is currently
Biologically Inspired Techniques with High working in the field of remote sensing
Resolution Satellite Image, Journal of the Indian and Geographic information system.
Society of Remote Sensing (Springer-Verlag), Vol. His research interests include
36(2), pp. 172 - 182, 2008. computer vision, digital image

Copyright © 2014 MECS I.J. Image, Graphics and Signal Processing, 2014, 10, 18-28
 28 Satellite Image Processing for Land Use and Land Cover Mapping

processing and Software Development. Mr. J Senthilnath, male, is a

research scholar at Indian institute of
science, Bangalore. He is currently
Dr S N Omkar, male, is the principal working in the field of remote sensing
research scientist at Indian institute of and Geographic information system.
science, Bangalore. He is currently His research interests include
working in the field of neural computer vision, digital image
networks, unmanned aerial vehicles, processing.
bio-mechanics, remote sensing and
Geographic information system.

How to cite this paper: Ashoka Vanjare, S.N. Omkar, J.Senthilnath,"Satellite Image Processing for Land Use and
Land Cover Mapping", IJIGSP, vol.6, no.10, pp.18-28, 2014.DOI: 10.5815/ijigsp.2014.10.03

Copyright © 2014 MECS I.J. Image, Graphics and Signal Processing, 2014, 10, 18-28

View publication stats