支持多源遥感数据格式的抽象库DFAL

针对当前遥感数据缺乏统一文件格式标准,多源遥感产品的生产使用涉及多种文件格式的现状,在研究多种遥感数据格式及对应库GDAL、HDF4、HDF5的基础上,参考国家气象局的NOAA AVHRR 1B格式标准,利用工厂模式,设计和实现了一个具有统一操作接口的多源遥感数据格式抽象库DFAL,在更高的抽象层次解决了GDAL和HDF4、HDF5库不能一致使用的问题。使用DFAL库进行不同文件格式的读写,专家无需了解这些文件格式的差别,以便从繁杂的遥感数据格式中解放出来,将更多精力投入到遥感科学研究的层面上。在DFAL库的基础上实现了两类常用遥感数据格式Geo TIFF和HDF5之间的转换工具,方便工程应用。DFAL库及转换工具在国家高技术研究发展计划《星机地综合定量遥感系统与应用示范》的多尺度按需定量遥感产品生产系统中进行了应用测试,取得了很好的实际效果,证明了DFAL库稳定性和可用性,并大大减少了多源数据协同使用时多种文件格式读写的工作量。

Abstraction library DFAL supported multisource remote sensing data formats

Remote sensing data lack a unified standard format, whereas multisource remote sensing products involve a variety of file formats based on the study of remote sensing data formats and the corresponding libraries GDAL, HDF4, HDF5, and NOAA AVHRR 1B format standard from China Meteorological Administration as reference. With the use of a factory model, a multisource remote sensing data format with a unified user interface is designed and implemented to obtain a better data format conversion tool between the GeoTIFF and HDF5 formats. Based on the study of various remote sensing image data file formats, the factory mode can not only shield the user from instantiating the details of different classes but also implement different types of operations to achieve a unified operational view, which is user-friendly. Given the lack of existing remote sensing data format libraries, GDAL, HDF5, and any format library cannot be used to read and write alien data formats consistently. Instead, the factory model is used, different remote sensing data formats are abstracted, and a new top-level architecture DFAL library is designed to solve the problem of reading and writing multisource remote sensing data formats. The multisource remote sensing data abstraction libraries DFAL are integrated into the underlying data IO for the quantitative production of a multisource remote sensing system. More than one multisource quantitative remote sensing production system is tested in the multitask parallel platform. The DFAL library causes no errors, and the usability and usefulness of the DFAL library have been proven. The DFAL library significantly simplifies the algorithm program for multisource quantitative remote sensing production, such as the production of multisource quantitative products mainly involving three different data formats that can be consistently read and written through the DFAL library and the IO code. Only one third of the logical workload is not used in the DFAL library. The multisource remote sensing data format abstraction library DFAL has generated a unified interface. Thus, remote sensing professionals can be liberated of the complex remote sensing data formats and can devote more energy to the study of the ontology of remote sensing. Future work can focus on three aspects of research and expansion, as below:(1) the remote sensing data formats are extended to ensure efficient processing by DFAL, and automatic formats for matching remote sensing data are achieved;(2) the projection transformation between different projections based on the DFAL library is supported, and the cooperative processing of multisource and multi projection data is applied;(3)the read and write operations are supported for multisource remote sensing time series data, which better serve the multisource and long time series applications of remote sensing.