Harvard University (Cambridge, MA, USA) researchers have developed a system that can take images of a face from one video source and map it onto another.
In the system, the video source that is to be inserted into a target video is first captured. The images of both the face in the source and target are then tracked and the source video is then warped in both space and time to align it to the target. Finally, the videos are blended together.
The system replaces all or part of the face in the target video with that from the source video. Source and target can have the same person or two different subjects. Either the source or the target can be existing footage, as long as the face poses are approximately the same.
According to Harvard researcher Kevin Dale, unlike previous systems, the approach uses inexpensive hardware and requires minimal user intervention.
The researchers believe that the video face replacement system will prove useful in both film and video editing applications.
More information is available on the Harvard web site.
-- By Dave Wilson, Senior Editor, Vision Systems Design
