基于多项式变换的遥感图像几何校正并行算法的研究与实现

Investigation and Realization of Parallel Algorithms for the Geometric Correction of Remote Sensing Images Based on Polynomial Transform

研究和实现高效的几何校正并行算法,能够有效地提高遥感图像预处理的工作效率。本文首先研究了基于多项式变换的几何校正算法,然后提出了新的并行几何校正算法。该算法通过局部边界计算达到延迟通信的目的,很好地解决了计算局部性问题;在机群系统上实现后,对多幅大小不同的G比特遥感数据集进行了测试,获得了正确的结果和良好的并行性能。

To study and realize effective parallel algorithms for geometrical correction can improve the job efficiency of remote sensing image preprocessing. Firstly, the geometrical correction based on polynomial transform is studied. Then a new parallel algorithm is proposed, which solves the problem of computing locality through calculating the local data edges to delay required communications. Finally, this parallel algorithm is realized on a cluster system, and several remote sensing images with the size of gigabytes are used to test the algorithm performance. The results show the validity and good efficiency of our algorithm.