欢迎访问中国科学院大学学报,今天是

中国科学院大学学报 ›› 2013, Vol. 30 ›› Issue (5): 676-681.DOI: 10.7523/j.issn.2095-6134.2013.05.016

• 信息与电子科学 • 上一篇    下一篇

基于CUDA的阈值迭代算法并行实现

耿旻明1,2, 蒋成龙1,2, 张冰尘1   

  1. 1. 中国科学院电子学研究所微波成像技术重点实验室, 北京 100190;
    2. 中国科学院大学, 北京 100190
  • 收稿日期:2012-06-15 修回日期:2013-01-21 发布日期:2013-09-15
  • 基金资助:
    国家973计划项目(2010CB731905)资助 

Parallel implementation of iterative shrinkage-thresholding algorithm via CUDA

GENG Min-Ming1,2, JIANG Cheng-Long1,2, ZHANG Bing-Chen1   

  1. 1. National Key Laboratory of Microwave Imaging Technology, Institute of Electronics, Chinese Academy of Sciences, Beijing 100190, China;
    2. University of Chinese Academy of Sciences, Beijing 100190, China
  • Received:2012-06-15 Revised:2013-01-21 Published:2013-09-15
  • Contact: 耿旻明,E-mail:gmm810@qq.com

摘要: 利用CUDA编程在GPU平台设计并行实现阈值的迭代算法,并应用于稀疏微波成像. 仿真实验结果表明,在正确重建信号的前提下,相对于常规的CPU串行计算,采用GPU并行处理能加快运算,提高成像速度.

关键词: 稀疏微波成像, 阈值迭代算法, 计算统一设备架构(CUDA), 并行处理

Abstract: We design and implement iterative shrinkage-thresholding algorithm (ISTA) on GPU via CUDA programming, and apply it in sparse microwave imaging. The simulation results show that, compared to CPU-based implementation, GPU-based implementation reconstructs correct signals at a faster computation speed.

Key words: sparse microwave imaging, iterative shrinkage-thresholding algorithm (ISTA), compute unified device architecture (CUDA), parallel processing

中图分类号: