Researchers at MIT and Google have developed a technique for removing unwanted reflections and obstructions from photos. Taking advantage of the parallax effect, this method could prove quite useful in locations where shooting behind glass or a fence is the only option.

Demonstrated in the (highly technical) video above, this method uses information gathered from a short burst of images captured as the photographer moves the camera. An algorithm searches for edges in the scene, making use of the parallax effect to determine which objects are near and far from the camera. Once identified, they can be separated and obstructions in front of the main subject can be removed.

This is a different approach to reflection removal than another we've recently seen published by MIT. That method identified and removed reflections by looking for the slightly offset double image you get when an image reflects from both the inside and outside surface of a piece of glass.

The latest work has the advantage that it works not only for reflections, but also for other occluding objects between the camera and the subject. In theory, the approach should extend to removing inconveniently placed telegraph poles or wires from images, so long as there is enough depth separation between the obstructing object and the background, and the photographer is able to move the camera enough to establish what's behind the obstruction.