• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2015, Vol. 37 ›› Issue (11): 2025-2029.

• 论文 • 上一篇    下一篇

基于OpenMP的包络对齐并行方法研究与实现

张佳佳1,姜卫东1,李宽2   

  1. (1.国防科学技术大学电子科学与工程学院,湖南 长沙 410073;2.国防科学技术大学计算机学院,湖南 长沙 410073)
  • 收稿日期:2015-08-21 修回日期:2015-10-20 出版日期:2015-11-25 发布日期:2015-11-25

Study and implementation of a parallel range alignment
algorithm based on OpenMP  

ZHANG Jiajia1,JIANG Weidong1,LI Kuan2   

  1. (1.College of Electronic Science and Engineering,National University of Defense Technology,Changsha 410073;
    2.College of Computer,National University of Defense Technology,Changsha 410073,China)
  • Received:2015-08-21 Revised:2015-10-20 Online:2015-11-25 Published:2015-11-25

摘要:

采用加矩形窗的积累互相关法和基于Fourier变换频域移位性质的最小熵法进行一维距离像包络对齐。针对包络对齐算法数据量大、复杂度高、运行时间长等缺点,提出一种应用于多核处理器的包络对齐并行算法。该方法利用OpenMP编译指导指令#pragma omp section和#pragma omp for对积累互相关算法和最小熵算法进行多线程并行优化。理论分析和仿真实验表明,该方法大大提升了算法的执行效率。

关键词: 包络对齐, OpenMP, 数据分块, 并行

Abstract:

By using the accumulated cross correlation algorithm based on rectangular window and the minimum entropy algorithm based on the frequency shifting property of the Fourier transform, we aligne a one dimensional range profile. Traditional range alignment algorithms suffer from large data, high complexity, long running time, etc. To overcome such problems, we propose a parallel range alignment algorithm for multicore processors. We use two compiling instructions "#pragma omp section” and "#pragma omp for” of OpenMP for the accumulated cross correlation algorithm and the minimum entropy algorithm to achieve multithreads parallel optimization. Theoretical analysis and experimental results show that the proposed method can increase the execution efficiency largely .

Key words: range alignment;OpenMP;data partition;parallel