Abstract: Traditional methods of processing remote sensing images could not effectively handle the mass daily production, mass pixel of single image, as well as the mass type and amount of objects. To solve the problem of image storage, we propose a structure for the storage of high-resolution remote sensing images in cloud computing environment, and expound the construction method based on MapReduce framework. We conducted experiments on large files of small image set in a Hadoop cluster and compared the image reading efficiency with that of traditional methods. The results proved that this storage structure has high scalability. Experiments also demonstrate this construction method has efficient reading/writing and processing ability.

Key words: cloud computing, high-resolution remote sensing image, storage structure, MapReduce, large files of small image sets, Hadoop