Energy-efficient high-performance SAR image geocoding with NVIDIA CARMA and its application in stereo radargrammetry

Author

Balz, Timo ; Lu Zhang ; Mingsheng Liao

Author_Institution

State Key Lab. of Inf. Eng. in Surveying, Wuhan Univ., Wuhan, China

fYear

2013

fDate

21-26 July 2013

Firstpage

2011

Lastpage

2014

Abstract

Energy is the major limiting factor for high-performance computing. Nowadays, modern Graphics Processing Units (GPU) use less power per floating-point operation. With the CUDA on ARM architecture, a low-power high performance architecture is available and we present first results on using this architecture for fast SAR geocoding and demonstrate the use of fast SAR geo-coding on GPU for a flexible SAR stereo-radargrammetry approach. We demonstrate the speed-up of SAR geo-coding with GPU implementations on an NVIDIA Tesla C2070 and the CARMA DevKit.

Keywords

geophysical image processing; graphics processing units; image coding; parallel architectures; radar imaging; stereo image processing; synthetic aperture radar; ARM architecture; CARMA DevKit; CUDA; GPU; NVIDIA Tesla C2070; energy-efficient high-performance SAR image geocoding; flexible SAR stereo-radargrammetry approach; floating-point operation; graphics processing unit; Abstracts; Graphics processing units; Indexes; Logic gates; GPU; SAR; geo-coding; radargrammetry;

fLanguage

English

Publisher

ieee

Conference_Titel

Geoscience and Remote Sensing Symposium (IGARSS), 2013 IEEE International

Conference_Location

Melbourne, VIC

ISSN

2153-6996

Print_ISBN

978-1-4799-1114-1

Type

conf

DOI

10.1109/IGARSS.2013.6723204

Filename

6723204