The search statistic used by PSS is
where 
is the data value in pixel 
, 
is its
variance, and 
is the point spread function (psf) amplitude
integrated over
pixel 
arising from a point source at position 
. The form of
this psf could in general be position-dependent. The pixels 
are
assumed to be regularly spaced and the summations in (13) extend
over all pixels for which p is significantly greater than zero (see
discussion below).
The numerator in equation (13) is just the weighted
cross-correlation of the
data with the psf, whilst the denominator is its standard deviation -- hence
is a measure of source significance. The search threshold should
obviously be set to avoid large numbers of spurious detections. For a
threshold value 
, one expects to get typically
spurious detections, where
is the area of the field being searched, 
is
the area under the psf, and 
is the integrated probability under
the upper tail of a unit normal distribution, beyond the value 
;
e.g. 
for
respectively.
This all assumes that any background has been removed
from the data d -- if not then it will contribute to the value of
and will increase the incidence of spurious detections.
