A new image processing technique has been applied to satellite photos of Mars [bbc.com], including the European Space Agency's failed Beagle 2 lander [wikipedia.org]. The processing has revealed a more definitive shape of the lander, which touched down in 2003. The photos, taken by NASA's Mars Reconnaissance orbiter (MRO), were analyzed using a "Super-Resolution Restoration" (SRR) technique by researchers from the University College London. The technique uses images taken from multiple angles and improved the resolution to about 5cm per pixel. The researchers also applied it to scenes involving the Curiosity and Spirit rovers, and anticipate using it to improve future Mars imagery:
The two scientists are now talking with Esa about using the technique to find a safe landing site for the agency's 2018 ExoMars rover; they are also keen to see it used on pictures from the new orbiting camera Europe has just sent to the Red Planet called CaSSIS [unibe.ch].
SRR is though a very computation-heavy approach. Using it on stacks to produce an image 2,000 pixels by 1,000 pixels can take three days on the MSSL Imaging Group's 14-blade (224-core) linux cluster. This should improve with time, and one driver could well be the desire to use super-resolution restoration on Earth-observation images.