deletions | additions       diff --git a/Results: Events.tex b/Results: Events.tex  index e3dc5e1..f59e072 100644  --- a/Results: Events.tex  +++ b/Results: Events.tex 
...
\section{Results \& Discussion}  \subsection{Event recognition and overlapping events} Typical single photon events detected with the EBCCD are shown in Fig~\ref{fig1}c,d. The central peak is high with small wings: during the diffusion of the electrons from the back of the sensor to the front, the charge spills over into adjacent pixels, although the pixel’s full well capacity is not reached. pixels.  Brighter, larger ion events are also detected, caused by a photoelectron hitting a gas molecule in the imperfect vacuum inside the EBCCD tube, leading to the gas molecule being ionised and accelerated towards the photocathode (Fig~\ref{fig1}c,d, top). The ion events cause problems with event recognition algorithms that find a threshold for each frame separately: the high brightness causes the threshold to be set too high and the photon events discarded as noise. The raw data was therefore preprocessed using ImageJ's tools by setting the intensity of all bright pixels in the ion events to a grey value slightly above the maximum intensity of the photons events. The ion events are then incorrectly localised as photon events, but due to the relatively rare occurrence of ion events compared to photon events (around 1 ion event per 600 photon events) this does not have a noticeable effect on the results.  For Photon counting images of  the MR processing, USAF test chart are shown in Fig~\ref{fig_results}. As  the enabling photon events are relatively dim compared to the high camera background noise, the sum  of MFA the frames without any processing produces a noisy image (Fig~\ref{fig_results}b), while centroiding  with one pixel accuracy removes camera the background and produces  a maximum clearer image (Fig~\ref{fig_results}c). Centroiding with 1/5-pixel accuracy (Fig~\ref{fig_results}d) recovers some  of2 molcules per fitting region introduces a bias into  the algorithm that it should anticipate detecting 2 photon events per fitting region. This can initially cause single photon events to be over-resolved into multiple events. The occasions where single events are over-resolved is counteracted resolution lost  by the remove duplicates function; electron diffusion in the sensor,  as shown in Fig~\ref{fig_results}e.  The mismatch between  the over-resolved detection markers are within event shape and  the same pixel, centroiding function can lead to fixed pattern noise which can be seen as bright and dark stripes in  the excess marker centroided image. The level of fixed pattern noise  can be removed whilst not affecting correctly-resolved overlapping events which quantified by  \begin{equation}  FPN = \frac{N_{max} - N_{min}}{N_{mean}} \times 100\%,  \end{equation}  where $N_{max}$, $N_{min}$, and $N_{mean}$  are typically further apart. the maximum, minimum and average number of counts in the 5$\times$5 array of subpixel positions, respectively. For a completely random distribution, this number should approach zero with increasing number of photons.