Thursday, March 28

New Algorithms Can Remove Photo Obstructions

Chain-link fences are usually considered more of a security measure than an aesthetic choice. If you’ve ever had an intrusive fence or a window glare in your photographs, a recent report from the Massachusetts Institute for Technology will be welcome news. Researchers at MIT and Google are developing a new algorithm, one that can separate the foreground and background of your image and remove unwanted obstructions.

The method works with a specific sequence of photos taken, similar to the panoramic feature on many smartphone cameras. The different angles provided by the multiple photos help to distinguish the foreground elements from the background elements, granting you the freedom to tweak them as needed.

Motion parallax is a phenomenon in which closer objects appear to move faster than those that are farther away, and it’s motion parallax that the researchers are using to advantage in the new algorithm.

“Since they’re moving differently, we can use that information to figure out there are two layers we’re actually looking at, and we remove one of them,” says Michael Rubinstein, a Google research scientist.

There are some early hurdles that will need to be overcome, however. The algorithm won’t work on subjects in motion, because it won’t work well in low light. Sporting games and other indoor performance events would also not be affected by the technology, and it may not work effectively if there are too many obstructions, like a chain link fence shot through a dirty window.

Rubenstein adds that while it is convenient for removing minor obstructions, “it’s not magic.”

The most effective use of the algorithm will be for pictures where there is a clearly lit and separated foreground, middleground, and background.

Leave a Reply

Your email address will not be published. Required fields are marked *