Very high resolution satellite data: New challenges in image analysis
Abstract
Early years of coming century will see a large number of satellites with very high spatial resolution reaching beyond 1 m in the visible range of electromagnetic spectrum. Such images will come very close to giving a ground-based view of a terrain with the exception that a ground-based view covers the entire optical range from 400 to 700 nm while satellite images will be wavelength-specific. Although the images will not surpass details observed by a human eye, they will, in principle, be comparable with aerial photographs obtained from the top of a skyscraper with suitable filters. This will generate enormous interest among users working in the area of coastal and urban studies where the rate at which features on the ground vary is much higher than that over oceans or other natural landscapes. Having very high resolution digital data over a landscape will however create new challenges in the field of atmospheric correction, ground registration, image processing and finally the image interpretation itself. Even the computer hardware presently in use for handling images with resolution from about 10 m to 1 km will become unsuitable for handling high resolution data. For example, one would need 100 images of 1 m resolution to cover an area on the ground that is covered by a single image of 10 m resolution. As a result, display of a single image of 1 m resolution would cease to give a synoptic view. Seaming smaller images to construct a synoptic view would make the image fairly big and unwieldy. Moreover, the excessive information made available through such efforts is more likely to shield the signal rather than enhance it. The present paper discusses some of the challenges posed by very high resolution satellite images and offers possible solutions around which the new developments in image analysis will have to evolve.
Collections
- Scholarly literature [9069]