High performance two-dimensional phase unwrapping on gpus

ISCAS OpenIR

	High performance two-dimensional phase unwrapping on gpus
	Wu, Zhenhua (1); Ma, Wenjing (1); Long, Guoping (1); Li, Yucheng (1); Tang, Qiuyan (2); Wang, Zhongjie (2)
	2014
会议名称	11th ACM International Conference on Computing Frontiers, CF 2014
会议日期	May 20, 2014 - May 22, 2014
会议地点	Cagliari, Italy
收录类别	EI
出版地	Association for Computing Machinery
ISBN	9781450328708
部门归属	(1) Lab. of Parallel Software and Computational Science, Institute of Software, University of Chinese, China; (2) Science and Technology on Integrated Information System Laboratory, Institute of Software, Chinese Academy of Sciences, China
摘要	Phase unwrapping is an important procedure in digital im-age and signal processing, and has been widely used in many elds, such as optical and microwave interferometry, mag-netic resonance imaging, synthetic aperture radar, adaptive optics. Phase unwrapping is a time consuming process with large amount of calculations and complicated data depen-dency. A number of algorithms with different features have been developed to solve this problem. Among all of them, Goldstein's algorithm is one of the most widely used algo-rithms, and has been included in some standard libraries (such as MATLAB). In this paper we propose an innovative implementation of Goldstein's algorithm on GPU. Several important approaches and optimizations are proposed for the GPU algorithm. For example, by introducing a local-matching step, we were able to parallelize the branchcut step effciently, getting much better performance than ex-isting work. With a cascaded propagation model, another important operation in the algorithm, floodfill, is able to make good use of the computing power of GPU. We tested our GPU algorithm on NVIDIA C2050 and K20 GPUs, and achieved speedup of up to 781 and 896 over the CPU imple-mentation respectively. To the best of our knowledge, this is the best performance of unwrap ever achieved on GPUs. © 2014 ACM.; Phase unwrapping is an important procedure in digital im-age and signal processing, and has been widely used in many elds, such as optical and microwave interferometry, mag-netic resonance imaging, synthetic aperture radar, adaptive optics. Phase unwrapping is a time consuming process with large amount of calculations and complicated data depen-dency. A number of algorithms with different features have been developed to solve this problem. Among all of them, Goldstein's algorithm is one of the most widely used algo-rithms, and has been included in some standard libraries (such as MATLAB). In this paper we propose an innovative implementation of Goldstein's algorithm on GPU. Several important approaches and optimizations are proposed for the GPU algorithm. For example, by introducing a local-matching step, we were able to parallelize the branchcut step effciently, getting much better performance than ex-isting work. With a cascaded propagation model, another important operation in the algorithm, floodfill, is able to make good use of the computing power of GPU. We tested our GPU algorithm on NVIDIA C2050 and K20 GPUs, and achieved speedup of up to 781 and 896 over the CPU imple-mentation respectively. To the best of our knowledge, this is the best performance of unwrap ever achieved on GPUs. © 2014 ACM.
语种	英语
内容类型	会议论文
URI标识	http://ir.iscas.ac.cn/handle/311060/16597
专题	中国科学院软件研究所
推荐引用方式 GB/T 7714	Wu, Zhenhua ,Ma, Wenjing ,Long, Guoping ,et al. High performance two-dimensional phase unwrapping on gpus[C]. Association for Computing Machinery,2014.