Title :
Three GPU-Based Parallel Schemes for SAR Back Projection Imaging Algorithm
Author :
Kebin Hu ; Xiaoling Zhang ; Wenjun Wu ; Jun Shi ; Shunjun Wei
Author_Institution :
Sch. of Electron. Eng., Univ. of Electron. Sci. & Technol. of China, Chengdu, China
Abstract :
Back projection (BP) algorithm has the ability of accurate imaging for synthetic aperture radar (SAR). However, the drawback of low computational efficiency limits the wide application of BP. In this paper, three different parallel schemes based on GPU for BP are presented. According to the features of serial structure of BP, the first two schemes are parallelized through mapping the pulse repetition intervals (PRIs) and pixels in the scene onto the parallel threads in GPU, respectively, another scheme combines the first two schemes together. In order to select the best scheme for application, we also provide a coarse criterion after detailed analysis of the three schemes from aspects of parallelization degree, memory occupancy and memory access efficiency. The simulation results are carried out and compared, which validates the efficacy of the three schemes.
Keywords :
graphics processing units; multi-threading; radar imaging; synthetic aperture radar; BP algorithm; GPU-based parallel schemes; PRI; SAR back projection imaging algorithm; coarse criterion; memory access efficiency; memory occupancy; parallel threads; parallelization degree; parallelized schemes; pixel mapping; pulse repetition interval mapping; serial structure features; synthetic aperture radar; Algorithm design and analysis; Azimuth; Graphics processing units; Imaging; Instruction sets; Memory management; Synthetic aperture radar; GPU; back projection algorithm; parallel processing; synthetic aperture radar;
Conference_Titel :
Computational Science and Engineering (CSE), 2014 IEEE 17th International Conference on
Conference_Location :
Chengdu
Print_ISBN :
978-1-4799-7980-6
DOI :
10.1109/CSE.2014.87