Title: Separating signal from background:

Abstract:
A complicating feature in certain astronomical data sets is that there is
contamination by foreground/background objects. We consider a sample of
stars from the dwarf spheroidal Sextans (with contamination), where some
stars are foreground stars, located along the line of sight. We develop an
algorithm for estimating parameters of the distribution sampled with
contamination. Our approach is based on the well-known "expectation
maximization" (EM) algorithm. Given models for both member and contaminant
populations, the EM algorithm iteratively evaluates the membership
probability of each discrete data point, then uses those probabilities to
update parameter estimates for member and contaminant distributions. We
also discuss some non-parametric extensions of this approach.