高光谱图像预处理的Matlab并行化研究  

Parallelization research on hyper-spectral image preprocessing based on parallel Matlab environment

作　　者：梁亮[1] 朱超[1] 杨捷[1] 吴素萍[1] 

机构地区：[1]宁夏大学数学计算机学院,宁夏银川750021

出　　处：《计算机工程与设计》2015年第8期2157-2161,2190,共6页Computer Engineering and Design

基　　金：国家科技支撑计划课题基金项目(2014BAD05B02);国家自然科学基金项目(60963004)

摘　　要：在高光谱图像处理中,针对预处理过程中计算耗时的问题,根据数据特征,应用Linux集群Matlab分布式并行处理技术,搭建Linux集群环境下Matlab分布式并行计算环境,在该环境下对高光谱图像预处理过程进行并行优化。通过实例进行验证,针对不同高光谱数据量在不同worker个数的情况进行测试,实验结果表明,该算法可高效完成高光谱图像预处理过程,对其消耗时间、加速比以及并行效率进行分析。Aiming at the time-consuming problem in the hyper spectral image pretreatment process, a distributed parallel compu- ting environment of Matlab under Linux cluster was built, so as to parallel and optimize the pretreatment process of hyper spectral image. Then experiment was performed to support this idea by testing different amounts of hyper spectral data and different numbers of workers. The experimental results show this parallel algorithm can improve the computing efficiency of hyper spectral image pretreatment process furthermore. The time-consumption, speedup and parallel efficiency of the parallel algorithm were also analyzed.

关 键 词：高光谱图像处理 LINUX集群 Matlab分布式 并行计算 并行优化 

分 类 号：TP302.7[自动化与计算机技术—计算机系统结构]