How to stack static pixels from frames to construct a clean background?

edit

I'm interested in combining the static components of multiple frames to generate to create an image without the dynamic contributors. This is a first step to perform accurate line recognition on a sports field. Does anyone have a good suggestion on how to script it in python?

Currently I'm using the absdiff to obtain the dynamic pixels and use them to create an inverted, slightly diluted mask that I slowly fill by using other frames.