Conclusions
This paper applied a straightforward and highly parallelized
mechanism to investigate parallel performance of five catego-
ries of typical algorithms in remote sensing based mapping on
two multi-core computers. According to these experimental
results, conclusions can be drawn that parallel computing on
a multi-core computer for these algorithms not only yields
high speedups but also efficiently leverages the computational
resources of the state-of-the-art hardware platforms. The
achieved performance of each algorithm on two multi-core
computers both with two
CPUs
is improved significantly. The
highest speedups of these algorithms range from 4.8× to 13.6×.
The parallel method adopted in this paper is quite easy to
implement and is seamlessly integrated with the five catego-
ries of typical algorithms. The method has a broad applicabil-
ity. Because these algorithms require reading and writing a
large amount of data and the component dealing with reading
and writing is less efficient than other components in a multi-
core computer, a good parallel strategy is that the computa-
tion and disk
I/O
operations occur simultaneously resulting in
decrease of overall runtime. The adopted parallel method not
only enables multiple blocks to be processed concurrently on
multiple cores but also implements the overlap between the
computational time and the time of reading and writing. Thus,
the parallel method is optimized for a multi-core computer.
Parallel performance on a multi-core computer is deter-
mined by these factors: the amount of required computation
per image size, the amount of required disk
I/O
per image
size, and the performance of the computing platform. The
larger the amount of required computation, the higher the
achieved maximum speedup. Hence, an algorithm requir-
ing a large amount of computation can be scaled. The large
amount of required disk
I/O
per image size is a key factor that
prevents improving the speedup to an even higher level for
some algorithms. In order to decrease the time of reading and
writing, high performance disk should be in high demand in
a computing platform. The number of
CPU
cores in the latest
computers with symmetrical multi-processing (
SMP
) and hy-
per-threading technology can meet the requirements of most
algorithms in remote sensing based mapping to some extent.
The testing results and contributions can guide ordinary
users in remote sensing community to promote processing
speed using desktop computing platforms. In the course of
parallel processing of typical algorithms in remote sensing
based mapping on a multi-core computer, optimal selections
are recommended as follows. The number of processors rang-
es from 8 to 12 for computers with high-frequency
CPU
(s) or
from 12 to 16 for computers with low-frequency
CPU
(s); high
performance disk, e.g., solid-state drive disk is in demand as
high as possible; and either Windows 7 OS or Linux OS is
appropriate.
Acknowledgments
This work was supported by the National Natural Science
Foundation of China under Grant No. 40901229, 863 Program
under Grant No. 2011AA120401, and the Basic Research
Fund of Chinese Academy of Surveying and Mapping under
Grant No. G7771413. Special thanks are given to the data
providers for the provision of the stereo data sets, especially
Euromap for the Cartosat-1 data. The authors thank the
anonymous reviewers for their valuable comments.
References
Adrov, V.N., M.A. Drakin, and A.Y. Sechin, 2012. High Performance
photogrammetric processing on computer clusters,
International
Archieves for Photogrammetry, Remote Sensing and Spatial
Information Sciences
, XXXIX-B4:109–112.
ASTRIUM, 2013. PIXEL FACTORY™, The Power of an Industrial
Solution in your Hands, URL:
/
files/pmedia/public/r367_9_geo_014-pixelfactory_en_4p
p.pdf
(last date accessed: 18 March 2015).
Bernabe, S., S. Sanchez, A. Plaza, S. Lopez, J.A. Benediktsson, and R.
Sarmiento, 2013. Hyperspectral unmixing on GPUs and multi-
core processors: A comparison,
IEEE Journal of Selected Topics in
Applied Earth Observations and Remote Sensing
, 6(3):1386–1398.
Campbell, N., 1996. The decorrelation stretch transformation,
International Journal of Remote Sensing
, 17(10):1939–1949.
Christophe, E., J. Michel, and J. Inglada, 2011. Remote sensing
processing: From multicore to GPU,
IEEE Journal of Selected
Topics in Applied Earth Observations and Remote Sensing
,
4(3):643–652.
Du, Q., N. Raksuntorn, A. Orduyilmaz, and L.M. Bruce, 2008.
Automatic registration and mosaicking for airborne multispectral
image sequences,
Photogrammetric Engineering & Remote
Sensing
, 74(2):169 -181.
Förstner, W., 1986. A feature based correspondence algorithm for
image matching,
International Archives of Photogrammetry and
Remote Sensing
, 26(3):150–166.
Fraser, C., G. Dial, and J. Grodecki, 2006. Sensor orientation via
RPCs,
ISPRS Journal of Photogrammetry and Remote Sensing
,
60(3):182–194.
Gillespie, A.R., A.B. Kahle, and R.E. Walker, 1986. Color
enhancement of highly correlated images - I. Decorrelation
and HSI contrast stretches,
Remote Sensing of Environment
,
20(3):209–235.
Grodecki, J., G. Dial, and J. Lutes, 2004. Mathematical model for
3D feature extraction from multiple satellite images described
by RPCs,
Proceedings of the 2004 ASPRS Annual Conference
,
Denver, Colorado.
Gropp, W., E. Lusk, N. Doss, and A. Skjellum, 1996. A high-
performance, portable implementation of the MPI message
passing interface standard,
Parallel Computing
, 22(6):789–828.
Han, S.H., J. Heo, H.G. Sohn, and K. Yu, 2009. Parallel processing
method for airborne laser scanning data using a PC cluster and a
virtual grid,
Sensors
, 9(4):2555–2573.
Lee, C.A., S.D. Gasster, A. Plaza, C.-I. Chang, and B. Huang, 2011.
Recent developments in high performance computing for remote
sensing: A review,
IEEE Journal of Selected Topics in Applied
Earth Observations and Remote Sensing
, 4(3):508–527.
Luo, W., B. Zhang, and X. Jia, 2012. New Improvements in parallel
implementation of N-FINDR algorithm,
IEEE Transactions on
Geoscience and Remote Sensing
, 50(10):3648–3659.
Nicolescu, C. and P. Jonker, 2002. A data and task parallel image
processing environment,
Parallel Computing
, 28(7):945–965.
Novak, K., 1992. Rectification of digital imagery,
Photogrammetric
Engineering & Remote Sensing
, 58(3):339–344.
PCI Geomatics, 2009. GeoImaging Accelerator Ortho Performance
Test Results, A PCI Geomatics White Paper, URL:
.
pcigeomatics.com/pdf/GXL_Ortho_Whitepaper.pdf
, (last date
accessed: 18 March 2015).
Plaza, A., D. Valencia, J. Plaza, and P. Martinez, 2006. Commodity
cluster-based parallel processing of hyperspectral imagery,
Journal of Parallel and Distributed Computing
, 66(3):345–358.
Plaza, A., Q. Du, Y.-L. Chang, and R.L. King, 2011. High performance
computing for hyperspectral remote sensing,
IEEE Journal of
Selected Topics in Applied Earth Observations and Remote
Sensing
, 4(3):528–544.
Plaza, A.J., 2008. Parallel techniques for information extraction
from hyperspectral imagery using heterogeneous networks of
workstations,
Journal of Parallel and Distributed Computing
,
68(1):93–111.
Reinartz, P., P. D’Angelo, T. Krauß, D. Poli, K. Jacobsen, and G.
Buyuksalih, 2010. Benchmarking and quality analysis of
DEM generated from high and very high resolution optical
stereo satellite data,
Proceedings of the 2010 Canadian
Geomatics Conference and Symposium of Commission I, ISPRS
Convergence in Geomatics - Shaping Canada’s Competitive
Landscape
, Calgary, Alberta, Canada.
384
May 2015
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING