Table 2. Computation Time of Our Method and Semi-
Automatic Method for Each Study Region
Image
A
B
C
D
E
Image Size
993×642 869×653 995×644 995×643 995×642
Proposed
method
0.087 0.078 0.083 0.086 0.081
Semi-automatic
1.15 1.029 1.171 1.170 1.166
Image
F
I
J
M N
Image Size
995×642 230×354 524×284 346×360 346×422
Proposed
method
0.080 0.024 0.048 0.036 0.08
Semi-automatic
1.165 0.04 0.076 0.066 0.16
There are mainly four differences and advances of pro-
posed method compared with other relevant researches: (1)
the landslide cases in our research area are more complicated,
and the size of our research area is larger; (2) compared with
many methods (Li
et al
., 2016; Martha, 2011), our method
has exactly one experienced parameter to detect landslides,
which has been proved effective in most cases as shown in
Figure 8. It is fast and more accurate than the state of the art
method (Li
et al
., 2016); (3) the method has a simple flow
chart and does not need post-processing or training samples
as done in many researches (Li
et al
., 2016; Martha, 2011),
which enhances the practical application of our method for
emergency response; (4) the input data for our method does
not depend on
SAR
or
DEM
as done in (Tsuchida
et al
., 2015),
and Landsat-8 images are publicly free. This availability
makes it more possible to be applied in more practical and
emergency cases.
The uncertainties of our method generally originate from
two domains. First, the quality of remote sensed images in
terms of lighting conditions and atmospheric environment
is a major concern for change detection method. Second, the
landslide background, especially for complicated background
with rocks, roads, and clouds, influence the accuracy of our
method to some extent. The proposed method may get im-
proved to further extract landslide by adding a classifier, and
use scale and lighting invariant features, such as SIFT (Ke and
Sukthankar, 2004).
Conclusions
This paper introduces a practical method in landslide map-
ping based on change detection. It first detects change of
inverse
NDVI
between pre- and post-landslide images, and
calculates saliency probability to detect landslides. The con-
cept “saliency “ is introduced in this paper to highlight the
landslide region, because landslide region commonly takes
a large contrast in spectral and spatial space, compared with
background objects, which share the similar characteristics of
ordinary saliency object in an image.
The experimental results of the proposed method have
demonstrated the strong efficiency and robustness in land-
slide detection, especially for the cases of complicated back-
ground in a large area.Our method is fast and simple to imple-
ment, and outperforms the state-of-the-art semi-automatic
method in landslide detection, in terms of both accuracy and
time consumption. Although the proposed method needs fur-
ther improvement to deal with the uncertainties from remote
sensed images and the various landslide background objects
in the future research, it can be practically used to provide
reliable landslide locations and areas for large-scale landslide
hazard assessment and emergency responses.
Acknowledgments
This study has been conducted with the support from the
National Science Foundation for Young Scientists of China
[grant number 41601451]; the Hundred Talents Program of
Chinese Academy of Sciences [grant number Y34004101A];
the Comparative Study on Global Environmental Change Us-
ing Remote Sensing Technology [grant number 41120114001];
and the National Natural Science Foundation of Major Inter-
national (regional) Collaborative Research Project, and High
Resolution Earth Observation Systems[grant number 14CNIC-
032079-32]. We are grateful for the NASA’s Earth observation
program in providing Landsat data from the website
http://
landsat.usgs.gov
and the offer of help as well. There are no
conflicts of interest among the researchers.
References
Akcay, H.G., and S. Aksoy, 2008. Automatic detection of geospatial
objects using multiple hierarchical segmentations,
IEEE
Transactions on Geoscience and Remote Sensing
, 46(7):2097-
2111.
Baldo, M., C. Bicocchi, U. Chiocchini, D. Giordan, and G. Lollino,
2009. LIDAR monitoring of mass wasting processes: The
Radicofani landslide, Province of Siena, Central Italy,
Geomorphology
, 105(3):193-201.
Bot, H.D.T., 2015. Nepal earthquake landslide locations, 30 June
2015, URL:
(last date
accessed: 21 March 2017).
Carrara, A., F. Guzzetti, M. Cardinali, and P. Reichenbach, 1999. Use
of GIS technology in the prediction and monitoring of landslide
hazard,
Natural Hazards
, 20(2-3):117-135.
Cheng, G., L. Guo, T. Zhao, J. Han, H. Li, and J. Fang, 2013. Automatic
landslide detection from remote-sensing imagery using a scene
classification method based on BoVW and pLSA,
International
Journal of Remote Sensing
, 34(1):45-59.
Colkesen, I., E.K. Sahin, and T. Kavzoglu, 2016. Susceptibility
mapping of shallow landslides using kernel-based Gaussian
process, support vector machines and logistic regression,
Journal
of African Earth Sciences
, 118:53-64.
Corominas, J., C. Van Westen, P. Frattini, L. Cascini, J.-P. Malet, S.
Fotopoulou, F. Catani, M. Van Den Eeckhaut, O. Mavrouli, and
F. Agliardi, 2014. Recommendations for the quantitative analysis
of landslide risk,
Bulletin of Engineering Geology and the
Environment
, 73(2):209-263.
Daniels, A.E., 2006. Incorporating domain knowledge and spatial
relationships into land cover classifications: A rule-based approach,
International Journal of Remote Sensing
, 27(14):2949-2975.
Di Martire, D., S. Tessitore, D. Brancato, M.G. Ciminelli, S. Costabile,
M. Costantini, G.V. Graziano, F. Minati, M. Ramondini, and D.
Calcaterra, 2016. Landslide detection
integrated system (LaDIS) based on in-situ and satellite SAR
interferometry measurements,
Catena
, 137:406-421.
Galli, M., F. Ardizzone, M. Cardinali, F. Guzzetti, and P. Reichenbach,
2008. Comparing landslide inventory maps,
Geomorphology
,
94(3):268-289.
Hervás, J., J.I. Barredo, P.L. Rosin, A. Pasuto, F. Mantovani, and S.
Silvano, 2003. Monitoring landslides from optical remotely
sensed imagery: The case history of Tessina landslide, Italy.
Geomorphology
, 54(1):63-75.
Holt, A.C., E.Y. Seto, T. Rivard, and P. Gong, 2009. Object-based
detection and classification of vehicles from high-resolution
aerial photography,
Photogrammetric Engineering & Remote
Sensing
, 75(7):871-880.
Huang, R., and X. Fan, 2013. The landslide story,
Nature Geoscience
,
6(5):325-326, URL:
/
static/63883857201051142734264/
(in Chinese).
Kääb, A., 2002. Monitoring high-mountain terrain deformation
from repeated air-and spaceborne optical data: Examples
using digital aerial imagery and ASTER data,
ISPRS Journal of
Photogrammetry and Remote Sensing
, 57(1):39-52.
362
May 2017
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
