An algorithm for reflection-free photographs
Researchers have found the solution that identifies and removes glass reflections in photographs
Most users trying to capture the beautiful view from the window of their hotel room or train have a hard time getting a photograph that does justice to the beauty outside. Professional photographers work around this by putting the camera as close to the window as possible and wrapping the lens in a dark cloth, which is then stuck to the window, for a tunnel effect. This prevents objects behind the camera being reflected off the glass.
An algorithm may now be able to help.
Four scientists at the Massachusetts Institute of Technology (MIT) in the US have created an algorithm that can distinguish between the actual scene and the reflections. They will present the algorithm at the Computer Vision and Pattern Recognition conference in June, in Massachusetts. YiChang Shih, the “first" author of the paper, is a PhD in computer science at MIT, co-authors Frédo Durand and Bill Freeman are professors of computer science and engineering, while Dilip Krishnan works at Google Research.
This is how it works—when multiple images are taken in quick succession, the algorithm identifies the identical reflections in each of those images. The software compares these photographs for any offset reflections from the glass, while the actual subject of the photograph remains constant. It divides images into blocks of 8x8 pixels, and calculates the correlation between each pixel per block. Such small blocks are useful because any massive colour changes are not likely in such a small amount of data being scanned, and the reflection of any external object will be visible around the edges.
This method doesn’t work if you want to eliminate reflections from a single photograph; you need to present the software with at least two-three images taken one after the other to enable it to work out the reflection elements.
At the moment, this algorithm is not available in any photography software or toolbox. But the potential is massive, simply because it can be integrated directly in cameras (particularly the professional DSLR cameras that allow a certain level of image-tweaking on the device itself), and will be extremely useful in software such as Adobe’s image-editing tools.
Consumers will have to wait till the official announcement next month to find out which photo-editing suites are likely to integrate this feature on the PC and mobiles.
Unlock a world of Benefits! From insightful newsletters to real-time stock tracking, breaking news and a personalized newsfeed – it's all here, just a click away! Login Now!