Derivation of accurate polarization information about an astronomical targe
t is a vital tool for investigation of astrophysical processes. Use of larg
e area detectors for imaging and spectroscopy has become commonplace, and f
requently such instruments offer a polarization capability. Processing of p
olarimetric data, however, is nontrivial, especially when the polarimeter i
s far from ideal. Here we present an overview of the analysis procedures ne
eded to properly process polarimetry data that comprise a series of images
of an object taken through a given set of polarizers, such as the imaging i
nstruments on the Hubble Space Telescope (HST). The analysis can also be us
ed for other types of polarization data, such as spectra. We consider only
linear polarization, not circular. The polarizers do not need to be perfect
polarizers, although it is important that their characteristics be well es
tablished. From an input data set of n intensities (or, equivalently, fluxe
s) and their errors, assumed independent between observations, correspondin
g to a set of observations through n polarizers (not necessarily identical
or perfect), we show how to derive the Stokes parameters and their covarian
ce matrix both for the special case of n =3 and for general n,
We then discuss how to derive higher level parameters such as polarization
degree and position angle and their associated uncertainties and indicate w
ays to "debias" the positive definite polarization degree. We present tests
of our analysis using Monte Carlo simulations. Finally, we show the achiev
able accuracy for various levels of polarization and signal-to-noise ratio
for typical cases, which should be useful for observation design. The techn
iques allow accurate recovery of polarization information from several of t
he instruments an board the HST as well as estimates of the uncertainties i
n the results.