摘要
空间应用需求对遥感数据处理的时间和精度提出新的要求,为了高速、高效地解决应用需求,提出一种基于通用模型的粗粒度遥感影像并行处理算法,采用"分块驱动"和"影像处理链驱动"两种策略进行遥感影像的并行算法设计,并对其进行MPI的实现。利用PCA融合算法在集群平台上进行并行性能测试,分析结果表明该算法在集群系统上获得了良好的视觉效果和近似线性的加速比,具有较好的扩展性和移植性。
The amounts of remote sensing data of global coverage will be grown exponentially.To study and achieve a fast and effective processing of these digital data with high accuracies has become a critical problem in remote sensing.This paper presents one coarse-grained method of parallel algorithm of remote sensing data processing based on general model.It inherits the modularization mechanism and uses both data partition and image processing chain to implement parallel processing with MPI library in cluster system.The experiments of PCA fusion algorithm is employed to test the efficiency and effectiveness of the parallel performance,which show that the new algorithm has the fine visual results,approximately linear speedup ratio and has fairly extensibility and transplantation.
出处
《遥感信息》
CSCD
2011年第3期14-18,120,共6页
Remote Sensing Information
基金
国家科技支撑计划基金(2008BAC34B07-04)
地理空间信息工程国家测绘局重点实验室基金(6101002)
关键词
通用模型
分块驱动
影像处理链
PCA融合
MPI
general model
data partition
image processing chain
PCA fusion algorithm
MPI
