Correction of electron microscope imaging parameters is a
problem that can be solved using automated procedures. However,
due to their lack of reliability, corrections are still often
performed manually. The purpose of this work is to implement
selected methods for the correction of imaging aberrations in a
scanning electron microscope. These methods are tested using
virtual motion in pre-scanned datasets that contain different
quality images of the same object. The tested method receives
at the input any image from the data set, and its goal is to
find the optimal image in this set in the shortest possible
time. Part of the work is also the evaluation of accuracy
and speed of the methods, evaluation of their advantages and
disadvantages. I will also compare the conventional approach
with methods using machine learning techniques.