SAR成像信号的并行处理算法研究

The Study of Parallel Processing Algorithm of SAR Imaging

对SAR并行成像处理进行了深入研究 ,在串行成像算法的基础上 ,提出了SAR并行成像处理的一种中粒度并行算法和一种混合FFT算法。中粒度并行算法针对SAR信号处理大数据量的特点 ,在系统内存有限的情况下 ,对任务划分进行优化 ,以提高系统处理性能。混合FFT算法将Cooley -TukeyFFT算法和WinogradFFT算法结合起来 ,使得FFT算法在通用平台上更高效地运行。在国产曙光系列并行机上的实验结果证明 :我们的研究工作是有成效的。

A medium grained parallel imaging algorithm based on the existing sequential algorithms are presented in this paper after the deep research on parallel processing of SAR imaging and a hybrid FFT algorithm is presented also. Taking the character of SAR data and limited system memory into accounts, the medium grained parallel imaging algorithm optimizes task partitioning to improve system processing power. Hybrid FFT algorithm combines Cooley-Tukey FFT with Winograd FFT so that FFT can run more efficiently on general computer platforms. The validity of our research is confirmed by the results of experiments on DAWNING parallel computer.