Fig. 10

Analyzes of the read-out noise \(\sigma _{read}^{2}\) and thermal noise variance \(\sigma _{therm}^{2}\) of gain normalized images under post-process binning of neighboring pixels. The lines represent the reconstruction using the Pearson correlation coefficients of Fig. 9 under binning. This is achieved following Eq. 26, which yields the variance of the binned image, as well as by utilizing Eqs. 27 and 22, which allow to calculate effect on the measured variance by the correlation of the binned image. The dots represent the results of the regression analyzes similar to Fig. 8. The colors indicate the respective detector quadrant and the colored shades depict the \(95\%\) confidence interval around the measured values. In (a), the read-out noise variances \(\sigma _{read}^{2}\) of a horizontal binning process are shown in dependence of the binning value H. In (b), the vertical binning in dependence of the binning value V is displayed and in (c), the diagonal binning is shown, where rows and columns HV were increased equally. In (d–f), the summations of the thermal noise \(\sigma _{therm}^{2}\) are shown using the same binning values as before. All reconstructions are in good agreement with the regression analyzes. However, the thermal noises seem to be underestimated by the regression analyzes that lead to the dots. As (d) and (e) show a linear increase of the thermal noise variance within tolerances, one would expect a quadratic increase for the binning in two directions. However, for higher diagonal binning values HV bigger than 10, the regression analyzes shows a discrepancy to this model.