使用OpenCL技术的影像快速畸变纠正方法在异构平台上的应用分析
发布时间：2018-10-05 07:11
【摘要】：针对海量遥感数据应用中日益显著的处理效率低下和计算瓶颈问题,基于通用计算机图形处理单元的编程开发使用OpenCL并行处理技术对遥感数据处理及其过程进行加速,旨在为遥感影像大数据处理提供一条更为高效的途径。在不同显卡平台上对影像畸变纠正实施并行处理,结果表明,OpenCL技术在提高影像畸变纠正的速度方面作用显著,可取得29.1倍的最高加速效果;与CUDA并行处理技术的交叉验证进一步凸显了OpenCL技术在异构平台上实施并行处理时所具有的通用性的优势。
[Abstract]:Aiming at the increasingly obvious problems of low processing efficiency and computing bottleneck in the application of massive remote sensing data, OpenCL parallel processing technology is used to accelerate the processing of remote sensing data and its process, which is based on the programming and development of general computer graphics processing unit. The aim is to provide a more efficient way for the processing of remote sensing image big data. Parallel processing of image distortion correction is carried out on different display card platforms. The results show that OpenCL technology plays a significant role in improving the speed of image distortion correction and can achieve the highest acceleration effect of 29.1 times. The cross-validation with CUDA parallel processing technology further highlights the common advantage of OpenCL technology in implementing parallel processing on heterogeneous platforms.
【作者单位】： 西南交通大学地球科学与环境工程学院;国家测绘地理信息局卫星测绘应用中心;
【基金】：国家自然科学基金(41474003) 四川省应急测绘与防灾减灾工程技术研究中心开放基金(K2015B007)资助
【分类号】：TP751
，
本文编号：2252479
[Abstract]:Aiming at the increasingly obvious problems of low processing efficiency and computing bottleneck in the application of massive remote sensing data, OpenCL parallel processing technology is used to accelerate the processing of remote sensing data and its process, which is based on the programming and development of general computer graphics processing unit. The aim is to provide a more efficient way for the processing of remote sensing image big data. Parallel processing of image distortion correction is carried out on different display card platforms. The results show that OpenCL technology plays a significant role in improving the speed of image distortion correction and can achieve the highest acceleration effect of 29.1 times. The cross-validation with CUDA parallel processing technology further highlights the common advantage of OpenCL technology in implementing parallel processing on heterogeneous platforms.
【作者单位】： 西南交通大学地球科学与环境工程学院;国家测绘地理信息局卫星测绘应用中心;
【基金】：国家自然科学基金(41474003) 四川省应急测绘与防灾减灾工程技术研究中心开放基金(K2015B007)资助
【分类号】：TP751
，
本文编号：2252479
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2252479.html
