摘要
传统的二维实阵列用于三维微波成像时,需要相对较高的脉冲重复频率才能切换发射阵元。针对这种情况,利用随机阵列来实现三维成像,目的是有效降低脉冲重复频率。进一步提出利用等效相位中心原理来实现任意随机阵列布局的可行性方案。然后,基于随机阵列,发现使用非均匀快速傅里叶变换(nonuniform fast Fourier transform,NUFFT)来实现后向投影(back projection,BP)算法的插值过程,能够用较少的运算量实现BP算法的高精度插值。最后,结合并行处理技术来实现NUFFT BP算法,结果使得BP算法的执行效率得到显著提高。
Traditional two dimensional （2D） antenna array requires high pulse repetition frequency （PRF） to switch transmitters, when it is applied to 313 microwave imaging. Therefore, 3B imaging is gotten by the random antenna array, as a result it reduces the requirement for PRF. Besides, a reasonable plan based on the phase center approximation （PCA） is proposed to implement the random antenna array of arbitrary geometry. The back projection （BP） algorithm interpolated by the nonuniform fast Fourier transform （NUFFT） is called NUFFT BP, and achieves high precision interpolation with relatively less computational complexity. Moreover, the NUFFT BP is further accelerated by the compute unified device architecture （CUDA） parallel computing technology, named CUBA NUFFT BP. It significantly improves the execution efficiency of the BP algorithm.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2015年第4期725-731,共7页
Systems Engineering and Electronics
基金
国家自然科学基金(61101170)
高等学校博士学科点专项科研基金(20110185110001)资助课题
关键词
随机阵列
非均匀快速傅里叶变换
并行处理
后向投影
random antenna array
nonuniform fast Fourier transform （NUFFT）
parallel computing
back projection （BP）
