Abstract
Registering an image to an earth coordinate system (geo-registration) is a crucial task in satellite image processing that facilitates the creation of image products (ortho-rectified imagery, rational polynomial coefficients, etc.). When the images are collected with a rolling-shutter camera in low-earth orbit, during imaging the satellite travels a significant distance relative to the image size, and rotates a significant amount relative to the angular size of a pixel. Therefore, georegistration requires a mathematical model that accounts for the motion and rotation of the satellite during imaging. This presentation discusses the development, calibration, and validation of a mathematical model using on-orbit data, along with estimated performance of satellite targeting and pointing provided by the model.
Geo-Registration of Satellite Imagery Captured in Low-Earth Orbit with a Rolling-Shutter Camera
Registering an image to an earth coordinate system (geo-registration) is a crucial task in satellite image processing that facilitates the creation of image products (ortho-rectified imagery, rational polynomial coefficients, etc.). When the images are collected with a rolling-shutter camera in low-earth orbit, during imaging the satellite travels a significant distance relative to the image size, and rotates a significant amount relative to the angular size of a pixel. Therefore, georegistration requires a mathematical model that accounts for the motion and rotation of the satellite during imaging. This presentation discusses the development, calibration, and validation of a mathematical model using on-orbit data, along with estimated performance of satellite targeting and pointing provided by the model.