无数据转置的并行SAR成像的研究和实现

Study and Implementation of Parallelized SAR Imaging Without Transposing Data Matrix

摘要: 目前比较成熟的SAR成像算法 ,无论是RD算法还是CS算法都需要有数据矩阵转置的操作 ,即完成数据从距离向到方位向的重新排序实现矩阵转置不仅需要有海量的存储空间 ,而且存储器I/O性能限制也使得完成转置操作需要耗费大量时间基于DSP系统 ,分别从算法和数据流的角度阐述两个免除数据转置的方法 ,并给出它们在DSP系统上的高效的并行实现实验表明这两个方法是可行且有效的 ,加快了SAR的成像处理

Abstract: For some conventional SAR imaging algorithms, such as Range Doppler algorithm or Chirp Scaling algorithm, matrix transposition is indispensable, which consumes lots of time and memory In this paper, two methods without matrix transposition are implemented in a parallel DSP system And experiment has proved the feasibility and effectiveness of these algorithms in speeding up SAR imaging