An ordinary camera could soon take photos of things that are out of sight, thanks to algorithms that interpret how light bounces off a wall.
鈥淣ormally, when light bounces off rough surfaces, like walls, it scrambles the scene into a messy blur,鈥� says at the University of Science and Technology of China, Hefei. 鈥淥ur goal was to 鈥榰nscramble鈥� that blur and recover the hidden scene. Think of it like turning a rough wall into a mirror.鈥�
The method involves mapping the geometry and reflectance of the wall surface by taking many images under different lighting conditions, so the researchers could predict how each bump and groove would distort reflected light. Once they had created a digital model of the surface, the team devised equations to reconstruct a hidden image from the scrambled light pattern.
Advertisement
Li and her colleagues have successfully demonstrated real-time imaging at 25 frames per second using an ordinary camera, like one found in a smartphone.
鈥淲hat makes this really exciting is that our method doesn鈥檛 require specialised camera equipment,鈥� says Li. 鈥淵our phone鈥檚 existing camera hardware is perfectly capable of capturing all the information we need. This means the technology could be implemented in smartphones, drones or other mobile devices without any hardware modifications.鈥�
at the University of Oxford says this approach improves on previous efforts to reconstruct images of hidden objects. 鈥淏y including a calibration step into their measurement, in which the properties of the wall are characterised, they can include the detailed spatial properties of the wall surface into the reconstruction, vastly improving resolution,鈥� she says.
A major limitation is that, so far, the technique can reconstruct images only from objects that emit their own light, such as LCD screens and smartphone displays. Most objects only reflect light from the environment, but this results in reduced imaging precision, Li says. The team hopes to address this by developing new algorithms for diverse surfaces.
Once the technology gets more advanced, it could have a wide range of applications, says Li. 鈥淔or law enforcement, it could provide officers with the ability to safely scan dangerous areas before entering,鈥� she says. 鈥淓ven autonomous vehicles could benefit 鈥� the system might help detect pedestrians hidden around blind corners, potentially preventing accidents.鈥�
Optica
