The role of Poisson's binomial distribution in the analysis of TEM images

Frank's observation that a TEM bright-field image acquired under non-stationary conditions can be modeled by the time integral of the standard TEM image model [J. Frank, Nachweis von objektbewegungen im lichtoptis- chen diffraktogramm von elektronenmikroskopischen auf- nahmen, Optik 30 (2) (1969) 171–180.] is re-derived here using counting statistics based on Poisson's binomial distribution. The approach yields a statistical image model that is suitable for image analysis and simulation.

► TEM images acquired under nonstationary conditions modeled using counting statistics.
► Poisson's Binomial distribution models counting process associated with TEM images.
► Resulting statistical image model generalizes existing ones based on Poisson models.
► The resulting statistical model useful for analysis and simulation of TEM images.