高光谱遥感影像分类算法并行处理设计与实现

探讨高光谱遥感影像分类算法处理遥感影像速度。通过光谱角度匹配（SAM）、光谱相关系数匹配（SCM）、信息散度匹配（SIDM）、光谱波形匹配（SWM）进行并行化改造设计，将改造的并行化算法应用到湖北大冶遥感影像数据分类处理中，结果表明并行化算法能够有效完成高光谱遥感影像分类，数据量增大，并行化处理速度加快，数据量为158×382×1092时， SAM 并行处理速度是串行处理速度的25．68倍、SCM 为25．41倍、SIDM 为17．55倍、SWM为23．68倍。并行分类算法处理遥感影像分类速度较串行分类算法处理快。

Parallel design and realization of classification algorithm of hyperspectral remote sensing image

This paper discusses the speed of the classification algorithm of hyperspectral remote sensing images.The classification algorithms of SAM,SCM,SIDM and SWM are designed by the parallel reconstruction method.The parallel reconstruction method is used to process the data of the hyperspectral remote sensing images which comes from Daye County in Hubei Province.The result shows that the parallel algorithm can effectively finish the classification of the images and the speed of the parallel processing is accelerated through increasing the amount of the image data.The parallel processing speed of SAM is 25.68 times,that of SCM is 25.41 times,that of SIDM is 17.55 times, and that of SWM is 23.68 times when the amount of data is 158×382×1092.The time of processing the remote sensing images by the parallel classification algorithm is shorter than that by the series classification algorithm.