If you’ve taken a picture through some sort of obstruction and haven’t been happy with the results, a group of researchers from MIT and Google feel your pain. The researchers detail their plan in “A Computational Approach for Obstruction-Free Photography.” Rather than capturing a single image, users take a short image sequence while slightly moving the camera, similar to taking a panoramic shot. An algorithm uses the differences that often exist in the relative position of the background and the obstructing elements from the camera and separates them based on their motions. See below for a before and after shot.
No comments:
Post a Comment