DocumentCode
2994632
Title
Research on CUDA-Based SIFT Registration of SAR Image
Author
Huang, Yang ; Liu, Jinshuo ; Tu, Meisheng ; Li, Siyuan ; Deng, Juan
Author_Institution
Comput. Sch., Wuhan Univ., Wuhan, China
fYear
2011
fDate
9-11 Dec. 2011
Firstpage
100
Lastpage
104
Abstract
This paper proposes an implementation and optimization of SIFT algorithm for SAR (Synthetic Aperture Radar) images. It improves the SIFT algorithm efficiency using the graphics processing unit (GPU) architecture based on Compute Unified Device Architecture (CUDA) framework, to attend the request of real-time remote sensing image processing. The experiments of large-size SAR image demonstrate that the algorithms achieve 57.8 times speedup.
Keywords
graphics processing units; image registration; parallel architectures; radar computing; radar imaging; remote sensing by radar; synthetic aperture radar; CUDA; SAR image registration; SIFT algorithm; compute unified device architecture; graphics processing unit; real-time remote sensing image processing; scale invariant feature transform; synthetic aperture radar; Algorithm design and analysis; Computer architecture; Feature extraction; Graphics processing unit; Image registration; Instruction sets; Optimization;
fLanguage
English
Publisher
ieee
Conference_Titel
Parallel Architectures, Algorithms and Programming (PAAP), 2011 Fourth International Symposium on
Conference_Location
Tianjin
Print_ISBN
978-1-4577-1808-3
Type
conf
DOI
10.1109/PAAP.2011.21
Filename
6128484
Link To Document