FROM WEATHER SATELLITES TO AERIAL IMAGING: BRINGING IMAGE NAVIGATION AND REGISTRATION DOWN TO EARTH

摘要

Image Navigation and Registration (INR) is an engineering term used with geostationary weather satellites. INR technology enables accurate determination of geo-locations ("navigation") for sensed objects. In weather satellite applications, image geometric distortions with respect to a standard fixed-grid projection are either compensated by onboard Image Motion Compensation (IMC) or by on-ground remapping ("registration"). Repeated acquisitions of the same image frame will then be registered to each other, which enables movie making. The use of geometric observations made through the aperture of the sensor itself is a general characteristic of such INR solutions. The current generation of Geostationary Operational Environmental Satellites (GOES) uses star senses and landmarks from which IMC model coefficients are determined. The next generation of GOES weather satellites will augment this capability with GPS from geostationary orbit. Current capabilities navigate and register imagery to the sub-pixel level. The INR problem is more challenging from an aerial platform as its flight dynamics are less predictable and consideration of the three-dimensional geometry of the scene is required. We describe our work in this area and the application of INR technology from GOES to the FalconScan commercial imaging venture. Low-cost FalconScan payloads fly on single engine airplanes, but they are well suited to UAVs once regulations permit such flights in national airspace. FalconScan processed almost 200 collections in 2011 for customers in the U.S. and Canada. Product mosaics are at 10cm to 50cm resolution and cover large areas that may include up to one billion pixels. Meter-level INR accuracies are achieved over these large survey spaces. We describe these products and some of the challenges in verification of their INR.