PSS performs a fit of its primary input to a model comprising background plus point spread function ( psf). At a fixed image position PSS finds that scale factor which multiplies the psf to optimise the match of the input data to the model -- this factor is the PSS flux. The units of this quantity are the same as the units of the input data, whatever they may be.
PSS can process the input data using two statistics, the Cash maximum likelihood statistic, or a correlation assuming gaussian errors. The former is intended for counts images where the errors are purely Poissonian, whereas the latter can be used on any data where the errors on each pixel are gaussian.
The PSS significance is a measure of the difference in the quality of the fit at the optimum flux compared to that at zero flux, expressed in normal deviates (``sigmas").