Pictures of the sky are unfortunately only 2-dimensional, so what we see is a projection of the 3-dimensional reality. Astronomers are naturally interested in the properties of the individual objects, and therefore need algorithms to estimate the properties of the isolated stars and galaxies, rather than just the assemblage.

I shall briefly discuss the Sloan Digital Sky Survey, which has generated about 10Tby of imaging data over the last 5 years, and then run over the characteristics of stars and galaxies that can be used to come up with an algorithm for separating overlapping images. I shall introduce the idealised algorithm, demonstrate that it works, and discuss various failure modes, along with ways in which the fundamental algorithm can be tuned to avoid at least some of them.