星载SAR图像几何校正并行算法研究

Research on Parallel Algorithm for Geometric Correction of Spaceborne SAR Image

并行计算是遥感图像快速处理的基本方法,而并行算法则是实现这种方法的关键之一.针对星载SAR图像几何校正中单幅图像的快速处理问题,首先分析了一种串行校正算法的可并行特征,随后分别设计了一种面向多CPU/核和一种面向GPU的全局并行校正算法,并利用2幅星载SAR图像测试了多种数据粒度下这2种算法的加速效能.实验结果表明,这2种并行算法可以大幅提高星载SAR图像的几何校正速度;CPU并行算法的性能主要与CPU/核心数目有关,而GPU并行算法则主要与数据粒度有关;后者具有很高的加速比,是几何校正的高效方法.

Parallel computing is a basic method to process remote sensing images quickly, and parallel algorithm is one of the key points to realize this method. Aiming at fast processing method of single image for geometric correction of spaceborne SAR images, this paper first analyzed the parallelizable characteristics of a serial correction algorithm, then designed a global parallel correction algorithm for multiple CPU/cores and GPU respectively. Finally, two spaceborne SAR images were used to test the speedup performance of the two algorithms under different data granularities. Experimental results show that the two parallel algorithms can greatly improve the speed of the geometric correction of spaceborne SAR image, and the performance of CPU parallel algorithm is mainly related to the number of CPU/cores, while the performance of GPU parallel algorithm is mainly related to the data granularity, and the latter has very high speedup ratio and is an efficient method for geometric correction.