面向应用层的遥感数据分块技术及其应用
Application Level Oriented Remote Sensing Data Segmentation Technology and Its Application

遥感是一种重要的地表观测技术，在许多领域有广泛应用。随着技术的不断进步，遥感技术已从传统低空间分辨率、多光谱向高空间分辨率、高光谱方向发展，使得遥感数据的体积过于庞大而无法一次性全部载入计算机内存进行处理。这种情况下，开展数据分块处理就显得尤为重要。当前针对遥感数据分块的模块主要在程序层级整合，且多数分块模块是商用的，这对应用层级的软件开发与集成并不友好。针对这一问题，本文提出一种面向应用层级别的分块机制，并利用IDL语言开发了针对遥感大数据文件自动分块和结果自动整合两个过程。在此基础上，以新疆吉木萨尔地区的航空高光谱条带数据为数据源，开展了分块、处理、整合实验，最终证明该机制和相关过程的可行性与实用性，不仅为后续开发应用层级别的遥感数据处理软件奠定基础，也可为开展数据并行处理和大数据网络传输提供帮助。
Remote sensing is an important surface observation technology that is widely used in various fields. With the development of the technology, the data of remote sensing has evolved from traditional low spatial and multispectral resolutions to high spatial and hyperspectral resolutions, which has made the volume of the data too large to be fully loaded into computer memory for processing at once. In this situation, the data segmentation is particularly important to data processing. Presently, modules for remote sensing data segmentation are mainly integrated during the programming process, and most of the modules are commercial, which constrains software development and in-tegration at the application level. This article presents a new segmentation mechanism to address the issue of application level data segmentation that has not been achieved by previous segmenta-tion modules. Two procedures, including the big data automatic segmentation procedure and the result automatic integration procedure, are developed using IDL language. The two procedures were used to do data segmentation experiments based on the airborne hyperspectral data of Ji-musaer area, and the results demonstrate the feasibility and practicality of the mechanism and related procedures. The achievements in this study not only contribute to the further development of the application-level integration software systems for remote sensing processing, but also help to carry out data parallel processing and big data network transmission.