摘要
针对遥感数据量大、处理时间长以及数据格式种类多的问题,基于国内外学者在遥感图像并行处理方面开展的工作,分析了GDAL开源库与OpenMP多线程程序设计库在遥感图像处理和提高处理效率中的作用,提出了遥感图像并行处理的图像分块方法以及在单机多核计算机环境下二者结合的方法;并通过实验测试了在单机多核计算机环境下不同并发线程数量对遥感图像并行处理效率的影响。结果表明,并行的加速效果与图像处理算法紧密相关,即便在并发线程不超过计算机核心数的情况下,也并非并发线程越多越好,不同的算法应结合实际情况选择合适的并发线程数量,以达到最高处理速度。
For the problems of large size of data and time-consuming and various data formats of remote sensing images,based on the analysis of domestic and foreign scholars parallel processing work in processing remote sensing image,we analyzed the function of GDAL and OpenMP library for improving the processing efficiency,and proposed a segmentation method for parallel processing by combining two libraries in single machine multicore computer environment.And then,we tested the efficiency of parallel processing of remote sensing images through experiments.The results show that the parallel acceleration effect is closely related with the image processing algorithm,and it is not the more threads the more better even in threads does not exceed the computer core number.Different algorithms should be combined with the actual situation to choose the appropriate number of concurrent threads,to achieve the highest processing speed.
出处
《地理空间信息》
2018年第11期55-59,125,共6页
Geospatial Information
基金
国家自然科学基金资助项目(41602333)
装备预先研究资助项目(41425050102)
关键词
并行处理
并行加速
遥感图像处理
GDAL
OPENMP
parallel processing
parallel processing speed up
remote sensing image processing
GDAL
OpenMP
