Abstract: In order to improve the storage and transmission efficiency of effective data, the massive satellite remote sensing images are constructed in units of multiple compressed sub blocks, and each image sub block is processed in real time cloud detection processing and counted in each scene image. At the same time, the satellite and orbit control data are used to locate the image scene center in real time, and the image scene center positioning information is matched with the template library of the region of interest present on the satellite. Combining the cloud proportion and matching results of the scene image, it is determined whether the foreground image is filled with clouds. Then conventional compression is performed on the non cloud scene and the scene of the regions of interest. For thick clouds and thin clouds in non interested areas, large compression ratio is adopted for compression. Experimental results show that this method which uses the cloud detection results and the importance of scene can guide the image compression, and significantly reduces the amount of compressed data in thick cloud regions and thin cloud area that is not interested.

Key words: optical telemetry satellite, cloud detection, real time positioning, intelligent compression