We performed aperture photometry on the restored images using a 2-pixel radius circular aperture. The small aperture size was selected to avoid confusion in the crowded regions. We evaluated both linearity of the measured stellar flux values and the RMS scatter in the results. Figs. 6 and 7 show the photometry results for one of the non-linear methods (1000 R-L/Snyder iterations) and one of the linear methods (minimum norm solution). The ratio of the measured counts divided by the actual counts in each star is plotted versus the magnitude of the star. The R-L/Snyder method shows significant non-linearity in the results for the fainter stars.
Table 3 shows the linearity of the results for each of the restoration methods for various ranges of magnitude and distance of the stars to their nearest neighbor. All methods overestimate the flux of stars having neighbors within 3 pixels. The measured flux values are significantly contaminated by the neighboring stars. For stars with separations greater than 3 pixels, all of the non-linear restoration methods show a non-linear behavior and give smaller median ratios for the fainter stars.
Table 4 shows the RMS scatter about the median ratios in Table 3.
None of the
methods give very good results for fainter stars with neighbors closer than 3
pixels. The MEM solution has slightly smaller RMS values than the other
non-linear methods, particularly for the brighter stars with a neighbor 3 to 6
pixels away. The hybrid method with has a lower RMS error
than the
linear methods while maintaining the linearity of the measured flux values.