Autocorrection of Frame Composition in Extra-Resolution Video

In the area of production of mass educational courses the lecture recording are performed. When editing these videos there is a need to use the different shot types. To facilitate the process of shooting and editing a presentation is recorded by one camera. The ultra high resolution is used for these purposes. Fortunately for lecture publishing on the internet the full HD resolution is quite enough. In that way it is possible to cut the close-up shots without loss of quality. It is simply enough to determine the area that should be cut, if the speaker does not move during the entire presentation. But the editing becomes quite laborious, if the motion takes the place in shot. In this work attempt are made to automate mentioned process. The proposal includes introduction to the study, the literature review and methodology description. The summary of results anticipated and conclusion are presented in the end.