RADIOMETRIC PROCESSING OF ADS IMAGERY: MOSAICKING OF LARGE IMAGE BLOCKS

Orthoimage mosaics presume radiometric adaptation of the individual images, which is referred to as relative radiometric normalization (RRN). This paper presents an in-house developed RRN approach that aims for optimal adaptation of large blocks of Leica ADS line-scanner imagery for subsequent mosaicking. It assumes orthorectified and radiometrically preprocessed input image data – with camera calibration, atmospheric and BRDF corrections applied as described in a companion paper – but still allows for the adaptation of bigger radiometric differences. The RRN model incorporates the adaptation of brightness and contrast, both varying throughout each image according to location-dependent polynomials. Correction parameters are derived from corresponding, invariant points within image overlaps; water areas are generally excluded to ensure smooth transitions over land. The computation is carried out as a block-wide least squares adjustment, combining both local adaptation of overlapping areas and global normalization of the mosaic. This RRN processing is used in North West Geomatics’ production since 2009 and has been applied to numerous ADS blocks. An example from TNRIS (Texas Natural Resources Information System) data is presented.