Direct georeferencing of airborne pushbroom images

Abstract

Direct georeferencing (DG) for orthoimage production based on collinearity equations has emerging as a standard photogrammetric operation. In this paper, the basic idea of DG is realized and refined with 19 self-calibration parameters for rectifying airborne pushbroom hyperspectral images collected using the intelligent spectral imaging system (ISIS) scanner. Two experimental ISIS images along with in-flight global positioning systems (GPS)/inertia measurement unit (IMU) data and 40 ground control points (GCPs) were used to solve these parameters by the iterative least squares method. By applying the proposed self-calibrated DG approach, the positional error of the GCPs is dramatically reduced to the pixel level from the 10-pixel level of in-flight DG-based rectification without self-calibration. Similar results were also found for the 30 check points, whose coordinates were either measured by GPS Real-Time Kinematic, or interpolated from 5-m to 20-m digital elevation model data, even when the six lens distortion parameters were omitted. It is demonstrated that the proposed self-calibrated DG approach promisingly improves the quality of georeferencing results by reducing the geometric distortions caused by instability of the platform, including misalignments in GPS/IMU, stabilizer effects on aircraft vibration and rotation, interior parameters of the sensor’s optical system, and variations in topography.