Operational airborne imaging spectrometers are generally either whiskbroom or pushbroom sensors. Image data acquired with such sensors are sometimes severly distorted by aircraft roll that occurred during image acquisition. This is especially so for data generated by sensors that are not carried in inertially stabilized mounts. The distortion is often so severe that the image data are not readily interpreted by eye. This is a great hinderance to quality control during image processing and analysis. If accurate sensor position and attitude data are acquired together with the image data, they may be used to rigorously remove geometric distortions from the image data. However, such ancillary data are often not collected. In the absence of such ancillary data, ROLLCOR may often remove enough roll distortion to make the image data interpretable by eye.
The only image transformation applied by ROLLCOR is the shifting of image lines by an integer number of pixels. Therefore, ROLLCOR does not introduce new image values, as may be the case with bilinear or cubic convolution resampling, in subsequent local (i.e., "pixel-by-pixel") image processing and analysis operations. Hyperspectral image processing and analysis is dominated by local operations, so the application of ROLLCOR may often be safely used as an early processing step in this context.
© PCI Geomatics Enterprises, Inc.®, 2026. All rights reserved.