基于OpenMP和OpenCV的归一化植被指数并行计算研究

Parallel Computing Reasearch of Normalized Difference Vegetation Index Based on OpenMP and OpenCV

归一化植被指数研究是遥感应用的主要领域,同时也是遥感学科的重要研究问题。随着我国遥感事业的进步,我国自主研制的高分辨率卫星先后陆续发射升空,加快并扩展了高分辨率卫星影像的应用范围。卫星每天产生的高分辨率遥感影像也呈现出指数级增长,达到Terabyte(TB)量级甚至Petabyte(PB)量级。然而传统串行的植被指数提取算法,已经不能及时并有效的从这些海量的遥感数据中提取植被指数。基于本文此以归一化植被指数(NDVI)算法为例,在高性能研究平台和并行计算技术的支持下优化NDVI提取算法,通过C++编程语言调用OpenMP和OpenCV库函数,实现了对NDVI快速提取的并行计算方法,并利用高分一号数据验证了NDVI并行提取算法的有效性。

The study of normalized difference vegetation index(NDVI)is the main field of remote sensing application,and also an important issue in the field of remote sensing.With the progress of remote sensing in China,the high resolution satellite developed by ourselves has been launched in succession,which has accelerated and expanded the application range of high resolution satellite images produces a large number of remote sensing images.High resolution remote sensing images produced by satellites every day also show exponential growth,reaching the order for Terabyte(TB)even Petabyte(PB)level.However,the traditional serial vegetation index extraction algorithm can’t extract the vegetation index from these massive remote sensing data in time and effectively.Based on this,this paper will use the normalized vegetation index(NDVI)algorithm as an example,Optimization of the NDVI algorithm,which based on the support of the high performance research platform and parallel computing technology,achieved the parallel computation method of NDVI fast extraction through C++programming language library function calling OpenMP and OpenCV and verified the validity of NDVI parallel extraction algorithm by GF-1 data.