Landslide Monitoring based on Computer Vision
Technology
Jiexian Zeng
Institute of Computer Vision
Nanchang Hangkong University
Nanchang, China
Email: zengjx58@163.com
Yonglong Yu
Institute of Computer Vision
Nanchang Hangkong University
Nanchang, China
Email:yylthm@126.com
Bo Tian
School of Information Engineering
Nanchang Hangkong University
Nanchang, China
Email: tianbo0319@163.com
Abstract—Landslide is a common geological disaster, which
seriously threats people's safety and property. The landslide
monitoring has great significance in reducing landslide loss. This
paper aims to design a new method for landslide monitoring with
computer vision technology. By this method, mark point should
firstly be located in regions where need to be monitored. Then
two sets of ordinary digital cameras are used to build stereo
vision system, achieve information acquisition in the region of
mark points. Finally, the change information before landslides
can be achieved through the identification, location, 3D
coordinate measurement and displacement judgment on mark
points. The experiment results have demonstrated that this
method has a high precision in landslide displacement monitoring.
It can be used in the real –time landslide monitoring.
Keywords—landslide monitoring, mark point recognition,
camera calibration, stereo vision system
I. INTRODUCTION
China is one of countries that suffer many geological
disasters, particularly the landslide. According to incomplete
statistics, more than 70 cities and 460 counties in China are
threatened and damaged by landslide, causing at least an
average annual loss of 1.5 billion to 2.3 billion yuan in China
[1]. Landslides are the ultimate expression of momentary
contiguous mountain slide. Nevertheless, landslide also has
some rules to follow. Long-term practical experience shows
that some obvious precursory phenomena usually appear
before big landslide come. For example, before the occurrence
of the landslide slope deformation is the most direct and most
easily captured information. Therefore, slope monitoring is
very meaningful, which is one of reliable ways for mountain
landslide hazard prediction.
The traditional method for landslide monitoring manual
measurement adopted by professionals is not only
time-consuming but also inaccurate- measuring. With the
development of science and technology, some new methods
are consistently applied to landslide monitoring. Reference [2]
shows the distributed optical fiber sensing technology of
BOTDR (Brillouin Optical Time Domain Reflectometer) of
Brillouin scattering is applied to monitor landslides. However,
deficiency of this method is that the laying of the sensing fiber
is difficult. Ref. [3] shows that InSAR (Interferometric
Synthetic Aperture radar) is applied to monitor landslides. This
method can be effective only when applied to the place where
there is sufficient scatterer and condition of relative low slope.
In [4], the use of GPS (Global Position System) technology for
landslide monitoring system can obtain the longitude, latitude,
and 3D (three-dimensional) coordinates. However, this method
has some problems, such as satellite signals can be blocked
easily in the mountainous areas, multipath effects are more
serious, causing negative influence on the measurement. Ref.
[5] with SIFT (Scale Invariant Feature Transform) feature
tracking method, determines the change in the number of
points to predict landslide, but feature extraction is difficult in
complex scenes, easily being mismatched. Ref. [6] shows that
a single camcorder is used to monitor landslides, a laser
projector project to two parallel plates in front of camera,
observing the laser spot offset to judge the movements of
mountain.
In this paper, computer vision method has been used to
achieve a landslide monitoring. First of all, layout point in the
monitoring area and collect information of monitoring area by
using a binocular camera system, then computer vision
technology is used to achieve the identification, positioning,
3D coordinate measurement and displacement judgment of
mark points, capturing the landslide change information before
slope.
II. M
ONITORING PRINCIPLES
The principle of monitoring method based on computer
vision technology is that using the camera for image
acquisition on the monitoring area, the application of computer
vision algorithms for image analysis, and landslides make
pre-judgment. As shown in Fig. 1, firstly lay artificial
landmarks in monitoring areas, then collect information of
monitoring area by using a binocular camera system and
capture the landslide change information before slope by the
identification positioning , 3D coordinate measurement and
displacement judgment of mark point. Finally, according to the
current situation judgment, the frequency of image acquisition
can be accelerated if the landslide is possible. At the same time,
the monitoring results will be transmitted to the monitoring
center through wireless transmission network in order to
achieve real-time monitoring and monitoring landslides.
This work was supported partially by the the National Natural Science
Foundation of China (Grant No. 61165011, 61263046), the Science and
Technology Support Program of Jiangxi province, China (Grant No.
20112BBG70092 & 20122BBF60072), Aerospace Science and Technology
Innovation Fund of China (Grant No. CASC201102).
____________________________________
978-1-4799-3197-2/14/$31.00 ©2014 IEEE