基于GPU的高光谱遥感岩矿信息快速提取方法研究
首发时间:2014-07-14
摘要:岩矿信息提取是高光谱遥感技术在地质探测方面的重要应用。随着高光谱遥感数据规模和计算复杂度的增大,岩矿信息提取面临着处理速度的挑战。针对此问题,该文利用GPU的并行计算优势对高光谱岩矿信息提取的核心步骤进行了并行优化设计,第一次提出了基于GPU的高光谱岩矿信息快速提取方法。该方法基于GPU并行编程平台,提出了基于GPU的岩矿信息快速提取并行设计模型。此外,针对高光谱岩矿信息提取的算法特点,提出了相应的性能优化策略,包括优化算法流程、减少访存次数和避免数据访问冲突。实验结果证明,提出的并行设计模型,与优化方法能够快速有效的进行岩矿信息提取,并且最大加速比达到了81倍。研究表明,GPU本身的硬件加速处理器十分适合高光谱遥感图像的处理,计算能力很强,花费时间代价也较少,能够满足实时处理的要求。
For information in English, please click here
The Research of Rock Mineral Information Extraction Fastly of Hyperspectral Remote Based on GPU
Abstract:Rock mineral information extraction is an important application of hyperspectral remote sensing technology in geological survey. With the increasing size and complexity of the hyperspectral remote sensing data, rock mineral information extraction faces the challenges of processing speed. To solve this problem, this paper proposes first the method of rock mineral information extraction quickly on GPU. With the advantages of parallel computing of the GPU, it optimizes the design to the core algorithms of rock mineral information extraction. The method raises the concurrent design model of rock mineral information extraction quickly on GPU parallel programming platform. In addition, for high spectral characteristics of rock and mineral information extraction algorithms, the method proposes some strategies of performance optimization, including optimization process, reduce memory access times and avoiding data access conflicts. Experimental results show that the model of parallel design, and optimization methods proposed can be carried out quickly and efficiently for information extraction of rock and mineral, and the maximum speedup achieved 81 times. Studies have shown that, GPU hardware acceleration processor that computing power is very strong and the price is also less time-consuming is very suitable for hyperspectral remote sensing image processing, which can meet the need of real-time hyperspectral remote sensing image processing.
Keywords: hyperspectral remotesensing mineral information extraction GPU parallel optimization
论文图表:
引用
No.4603309988270140****
同行评议
共计0人参与
勘误表
基于GPU的高光谱遥感岩矿信息快速提取方法研究
评论
全部评论0/1000