<i>In Situ</i> Mosaic Brightness Correction
- Sunday, 01 April 2012
In situ missions typically have pointable, mast-mounted cameras, which are capable of taking panoramic mosaics comprised of many individual frames. These frames are mosaicked together. While the mosaic software applies radiometric correction to the images, in many cases brightness/contrast seams still exist between frames. This is largely due to errors in the radiometric correction, and the absence of correction for photometric effects in the mosaic processing chain. The software analyzes the overlaps between adjacent frames in the mosaic and determines correction factors for each image in an attempt to reduce or eliminate these brightness seams.Two related methods of correcting brightness differences at seams between frames in a mosaic of in situ images work on the same general principle. The overlapping areas between adjacent frames in a mosaic are analyzed, and statistics are gathered. These statistics are then used in a bundle-adjustment style procedure to derive correction parameters for each image that minimize the brightness seams across the mosaic.
The older method consists of two programs: marsint, which gathers overlap statistics, and marsbias, which determines correction parameters. The newer system adds additional capabilities, including simultaneous brightness and contrast correction, better overlap statistics, improved image labels, and standard XML file formats. The overlap analysis functionality is embedded in the mosaic program marsmap, while the correction parameters are determined by the program marsbrt.
In both cases, the correction parameters are input to the mosaic program of interest (marsmap, marsmos, marsmcauley) to be applied to the mosaic. Correction parameters are constant additive or multiplicative factors applied to the entire input image; no nonlinear corrections are applied.
The software is part of the OPGS (Operational Product Generation Subsystem) software suite. While the algorithms behind this suite are not particularly unique, what makes the programs useful is their integration into the larger in situ image processing system via the PIG library and the mosaic programs. They work directly with space in situ data, understanding the appropriate image metadata fields and updating them properly.