GPU加速下的三维快速分解后向投影SAS成像算法
Three-dimensional fast factorized back projection SAS imaging algorithm accelerated by GPU
陶鸿博 1张东升 2黄勇3
作者信息
- 1. 中国科学院声学研究所海洋声学技术实验室,北京 100190;中国科学院大学电子电气与通信工程学院,北京 100049
- 2. 中国科学院声学研究所海洋声学技术实验室,北京 100190
- 3. 浙江大学宁波"五位一体"校区教育发展中心,浙江宁波 315100
- 折叠
摘要
后向投影(back projection,BP)算法是一种精确的时域成像算法,但BP算法的计算复杂度高,难以实现实时性成像,特别是在考虑三维成像时,BP算法的计算复杂度会进一步增加.提出一种应用在合成孔径声纳(synthetic aperture sonar,SAS)上的三维快速分解BP(fast factorized BP,FFBP)成像算法,并利用图形处理器(graphics processing unit,GPU)加速三维FFBP算法.经过对点目标的测试,计算时间从原本的263 s降低到了2.3 s,解决了 SAS中的三维成像实时性问题.同时,验证了所提算法在非理想航迹下的成像效果.结果表明,在添加幅度不超过0.1 m(一个波长以内)的正弦扰动时,所提算法对点目标仍有良好的聚焦效果.
Abstract
The back projection(BP)algorithm is an accurate time-domain imaging algorithm,but the high computational complexity of the BP algorithm makes it difficult to achieve real-time imaging.Especially when three-dimensional imaging is considered,the computational complexity of the BP algorithm will further increase.The paper proposes a three-dimensional fast factorized BP(FFBP)imaging algorithm applied to synthetic aperture sonar(SAS),and accelerates the three-dimensional FFBP algorithm using a graphics processing unit(GPU).After testing on point targets,the calculation time is reduced from the original 263 s to 2.3 s,solving the real-time imaging problem in SAS.At the same time,the imaging effect of the proposed algorithm under non-ideal trajectories is verified.The results show that when a sinusoidal disturbance with an amplitude not exceeding 0.1 m(within one wavelength)is added,there is still a good focusing effect on the point target.
关键词
快速分解后向投影/并行计算/图形处理器/合成孔径声纳/三维成像Key words
fast factorized back projection(FFBP)/parallel computing/graphics processing unit(GPU)/synthetic aperture sonar(SAS)/three-dimensional imaging引用本文复制引用
基金项目
中国科学院重点部署项目(KGFZD-145-22-06-02)
出版年
2024
NETL
NSTL
万方数据