Automatic colour/contrast adjustment
Automatic colour/contrast adjustment in large image mosaics
Operational use of satellite images often requires the production of area covers consisting of large numbers of individual scenes. In some cases it is advantageous to work with the individual scenes in the subsequent analysis, but in other cases there is a need to treat the full cover as a single image. Alternatively there may be a requirement to use other divisions than the scene boundaries, such as map sheets or administrative areas. The latter need calls for making a mosaic image from the dataset before sectioning into the required spatial entities.
When making a mosaic, it is important that the appearance (i.e. colour and contrast) is consistent between the scenes composing in the mosaic. Differences in appearance are, however, almost always present. Acquisitions at different dates with different atmospheric and illumination conditions introduce radiometric variations that can be adjusted to a large extent. Seasonal differences in ground cover are more difficult to compensate for. The contrasting characteristic of different sensors that may be used in the mosaic is another source of radiometric mismatch.
The radiometric matching of scenes in a mosaic is usually a manual task, where interactive colour/contrast balancing between scenes is performed visually. This gives very good results when dealing with only a few scenes. But with increasing numbers of scenes, this approach rapidly becomes extremely difficult or impossible.
The development of an automatic method for adjusting the colour and contrast of satellite images participating in a large mosaic is proposed. The automation will apply both to methods for measuring relative radiometric parameters in image overlaps in a robust way, and to adjusting the parameters for all participating scenes in a least-squares sense. The result will eliminate the present need for time-consuming manual measurements, and provide better quality in large mosaics. It will provide a possibility for fully automatic image mosaic production.
The developed methods will be implemented within Spacemetric’s production software, including an installation within the SIP/Ortho production environment. A successful implementation will give Metria and other producers of image mosaics more cost effective tools for the production of large image mosaics.
Torbjörn Westin, Spacemetric AB, tw(a)spacemetric.se
Jörgen Forsgren, Metria, jorgen.forsgren(a)lm.se
25 maj 2009