His final image is accurate provided that the exposure times for each of the three images is the same.
The only reason he might have something wrong is if the exposure times are much different between the images. This is very possible actually because taking a picture of the red martian surface will mean a very low exposure time through a red filter. The other two filters will require much more time to build up a good signal when looking at the surface, so this would allow a disproportionately large signal in those two bands from the sky.
I can't find any info on the exposure times though...