![](https://csdnimg.cn/release/download_crawler_static/18140795/bg1.jpg)
Application of image classification technology in mangrove information extraction
Shulei Wu
1
, Liya Dong
1
, Huandong Chen
1,*
1
College of Information Science and Technology
Hainan Normal Univers ity
HaiKou, China
E-mail: wus hulei@neteas e.com
Jinmei Zhan
2
2
Department of Information and Technology
Qiongtai Teachers College
HaiKou, China
* Corresponding author e-mail:
ch_huandong@163.com
Abstract— As the particularity of the mangrove ecosystem, it is
difficult to do field investigation, which consumes much time,
and the obtained data is not real-time. The technology of
satellite remote sensing plays an important role in the detection
of the ecological system. Some methods, such as band
combination, expert classification and fuzzy classification, have
been widely used in recent years. This paper explores the
application of the image classification technology in mangrove
information extraction based on RGB color only, and discusses
the method of parallelepiped and Meanshift clustering
respectively. According to the different image characteristics,
the image classification technology based on RGB color only
can distinguish the different research contents in remote
sensing image, which uses a computer to analyze the target
object quantitatively and classifies each image pixel or region
into several categories.
Keywords-image classification technology; satellite remote
sensing; parallelepiped method; Meanshift clustering method
I. INTRODUCT ION
Mangrove is an evergreen and small tree community and
grows in the tropical and subtropical coastal intertidal
zone[1]. As the high density of vegetation and the special
growth environment, it is difficult to do field investigation
and lots of effort and money need to be input[2]. Secondly,
mangrove forest land is flat, it is difficult to locate accurately
and contour extraction by the traditional forest resources
survey method. Therefore, at present, satellite re mote s ens ing
is used for research on mangrove at home and abroad mostly.
The technology of satellite remote s ens ing , which has the
advantages of large-area and single-time detection, can make
us obtain the remote sensing data of the same area
periodically. And it can also be a shortcut to the research of
the ecological and dynamic change of mangrove[3-6].
This paper explores the application of image
classification technology in mangrove information extraction
based on RGB color, and studies the method of
parallelepiped, Meanshift clustering method.
II. O
VERVIEW OF INTEREST AREA
Hainan Dongzhaigang National Nature
Reserve(longitude 110e32Ą110e37Ą, north latitude
19 e51Ą20e 1 Ą),located in the northeast of Hainan
Island, between Haikou City and Wenchang City, is the first
National Nature Reserve of our country established in 1980
in order to protect mangrove forests. The mangrove area is
1578.203 hectares. And the species in the mangrove area are
mainly Rhizophora stylosa, Bruguiera sexangula, Ceriops
tagal, Bruguiera gymnorrhiza, Kandeliacandel and
Aegicerascorniculatum[7]. The image of interest area is
shown in figure 1.
Figure 1. Image of interest area.
III. IMAGE CLASSIFICAT ION METHOD
Based on the image of interest area, we simulate the
artificial classification method. First, we consider it contains
water, large vegetation and the interference region. The
water area can be classified into the sea water and river. The
sea water can be classified into the offshore and high sea
further. The river can be classified into center river water and
boundary water further. The large vegetation area can be
classified into mangrove, farmland and other vegetation.
Mangrove area can be classified into the general, the sparse
and the dense area. According to the above, the classifier can
be constructed by using parallelepiped method[8].
2015 11th International Conference on Computational Intelligence and Security
978-1-4673-8660-9/15 $31.00 © 2015 IEEE
DOI 10.1109/CIS.2015.48
167
2015 11th International Conference on Computational Intelligence and Security
978-1-4673-8660-9/15 $31.00 © 2015 IEEE
DOI 10.1109/CIS.2015.48
167
2015 11th International Conference on Computational Intelligence and Security
978-1-4673-8660-9/15 $31.00 © 2015 IEEE
DOI 10.1109/CIS.2015.48
167
2015 11th International Conference on Computational Intelligence and Security
978-1-4673-8660-9/15 $31.00 © 2015 IEEE
DOI 10.1109/CIS.2015.48
167
2015 11th International Conference on Computational Intelligence and Security
978-1-4673-8660-9/15 $31.00 © 2015 IEEE
DOI 10.1109/CIS.2015.48
167
2015 11th International Conference on Computational Intelligence and Security
978-1-4673-8660-9/15 $31.00 © 2015 IEEE
DOI 10.1109/CIS.2015.48
167