【摘要】 最小噪声分离变换(MNF)是高光谱遥感影像分类中特征提取和去除噪声的有效方法.MNF算法涉及大量的矩阵运算,在实际工程的海量数据处理中存在计算时间长的问题.在分析MNF算法原理的基础上,运用图形处理单元(GPU)并行框架对该算法进行优化,并通过不同大小的高光谱遥感数据进行计算和分析.结果表明,随着影像数据量的递增,采用并行计算方式的提速比呈明显上升趋势,说明GPU并行方式对于计算密集型的大数据量处理具有良好的提速效果,为解决海量高光谱遥感数据处理速度慢的问题提供了思路.更多还原

【Abstract】 The most typical characteristic of hyperspectral remote sensing is the gigantic quantity of the data which is massive in practice.The modeling algorithm is so large that the traditional serial algorithm costs an extremely large compute time.This paper is research in the parallel processing technique that used in hyperspectral remote sensing data processing.Then it brings forward a basic flow of MNF parallel algorithmic based on GPU.At last,it makes a system experiment,the experiment results prove the validity of the algorithm.更多还原