数据密集型计算的遥感图像预处理方法

针对大数据时代，数据密集型计算已经成为国内外的一个研究热点. 遥感数据具有多源化、海量化特点，是名副其实的大数据. 研究适用于遥感影像自动化、业务化处理的数据密集型计算方法，是目前遥感应用技术面临的挑战所面临的挑战，本文提出了一种基于数据密集型计算的遥感图像处理方法. 在文中，首先围绕遥感数据自动化、业务化预处理等问题，深入调查和分析了国内外研究现状，进而介绍了系统体系结构，通过工作流灵活组织多种算法模型协同工作，设计以“5并行1加速”的计算体系解决数据密集型的遥感图像预处理，并通过产品生产实例对其性能进行测试. 结果表明，该系统在保证处理精度的前提下，大大提高了遥感大数据预处理的效率.

Remote Sensing Image Preprocessing Method Based on the Data Intensive Computing

In the era of big data, the research on data-intensive computing is becoming more and more popular both at home and abroad. As a typical branch of big data, remote sensing data is characterized both by the variety of the data sources and the huge data quantity. One of the biggest challenges facing the remote sensing application is how to find out a data-intensive computing method which aims at the automation of the business processions of remote sensing images. In this paper, a new data-intensive computing method for the procession of remote sensing images is proposed. After a deeply study focusing on the automation of the business processions of remote sensing data, a new systematic architecture using workflow is introduced which can coordinate the work among different algorithm models. In addition, in the pre-processing of the remote sensing images, a new computing architecture with five different types of parallelism and a stage of acceleration is also adopted. The computing method proposed in this paper has been tested in many products in real production environment in order to testify its effectiveness. The results show a significant improvement on the efficiency of the pre-processing of remote sensing data in the condition of ensuring the processing precision.