We present a method to directly visualise a statistical analysis of skyrmion defect alignment at grain boundaries in the skyrmion host Cu2OSeO3. Using Lorentz transmission electron microscopy, we collected large data sets with several hundreds of frames containing skyrmion lattices with grain boundaries in them. To address the behaviour of strings of dislocations in these grain boundaries, we developed an algorithm to automatically extract and classify strings of dislocations separating the grains. This way we circumvent the problem of having to create configurations with well-defined relative grain orientations by performing a statistical analysis on a dynamically rearranging image sequence. With this statistical method, we are able to experimentally extract the relationship between grain boundary alignment and defect spacing and find an agreement with geometric expectations. The algorithms used can be extended to other types of lattices such as Abrikosov lattices or colloidal systems in optical microscopy.