In the world of remote sensing, everything is big. The datasets provided by your preferred image supplier are shipped to you on one or many big disks. To view these image datasets you need a powerful workstation with large amounts of CPU, large amounts of RAM, and always a very good graphics card. Then, when you come to process the imagery, unless you are doing a small area analysis, it takes a long time as each pixel has to be analysed.
In the past our satellite images have had moderate resolution. Landsat presented us with 30m pixels, SPOT 6m pixels and WorldView-2 2m pixels. With the release of WorldView-3, the resolution has again increased. We can acquire multi-spectral satellite imagery at 1m resolution. Initially, this has been seen as a great advancement in remote sensing technology, but with the great gains there is a downside. The size of the disk required to supply the same size image as WorldView-2 has increased exponentially, and the time to process these new images has also increased.
This methodology of bringing the image to the application to process is becoming less viable. It is becoming more expensive in terms of the hardware required, and time-consuming for the image processing. The thought practice now is instead of bringing the image to the processor, why not take the processor to the image. Enter the world of the Cloud.
All the major image providers are now offering imagery as web services. Taking a look at the ArcGIS Marketplace, Airbus, RaidEye and DigitalGlobe are all supplying imagery as ArcGIS image services in full spectral resolution, so no longer is there a need to order imagery from these providers and have it shipped – you can simply order and have it put in the archive when it becomes available, then access it as an image service directly in your desktop.
This is fine if you want to just view it, but what about processing. In many situations, processing the full spectral image service on the desktop is not viable. Exelis has come to the party with the release ENVI Services Engine (ESE). Essentially this is image processing in the Cloud on-the-fly as you need it against image services.
If you are performing repetitive tasks on large datasets such as atmospheric correction, orthorectification, dark body subtraction and then change detection, ESE allows you to string all these processes together, process the image services using the power of the Cloud and put the results out as an image service that can be consumed by geospatial applications.
No longer is there the requirement to have grunty workstations to perform image processing – anyone with a browser can access ESE and execute image processing tasks. Exelis has released a very good webinar which describes this process.
To take this one step further, Exelis and DigitalGlobe have an agreement to place ESE in DigitalGlobe’s archive. For the consumer, this means no need to purchase Cloud resources for either imagery or ESE processing. More details on this offering are coming but for now the agreement can be found on the Exelis website.
All this means is image processing is becoming easier for the consumer. The move to software-as-a-service has started, and I for one thought this could not happen in the image processing space.
Happy image processing.