Advanced Geosynchronous Studies Imager (AGSI): image navigation and registration (INR) system

摘要

Abstract: Accurate and automatic image navigation and registration (INR) of remotely sensed data will be an essential element of future NASA satellite observation systems. INR describes the process by which geographic locations of the image pixels are computed and successive images from the same sensor are aligned to each other over time. For sensors such as the Advanced Geosynchronous Studies Imager (AGSI), a number of distortions prevent successive images from being perfectly registered to each other or to a fixed coordinate system. Most distortions in such images are the combined effects of sensor operation, satellite orbit and attitude, and atmospheric and terrain effects. These distortions are usually corrected by two methods; systematic correction, which relies on image acquisition models taking into account satellite orbit and attitude, sensor characteristics, platform/sensor relationship, and terrain models, and precision correction, which is feature-based, starting from the result of the systematic correction, and refining the geolocation or relative registration to subpixel precision. This paper describes the AGSI INR requirements and concepts, the image navigation model, a description of some potential precision correction methods utilizing edge and wavelet features, and a study of all the different error sources. The issues of swath-to-swath correlation and channel-to- channel coregistration are also described. !19