Seeing the invisible: a camera that looks around corners

Citation

Bogue, R. (2011), "Seeing the invisible: a camera that looks around corners", Sensor Review, Vol. 31 No. 3. https://doi.org/10.1108/sr.2011.08731caa.002

Seeing the invisible: a camera that looks around corners

Article Type: Mini features From: Sensor Review, Volume 31, Issue 3

A research group from MIT have recently announced the seemingly impossible: seeing around corners. The new technique is dubbed “femtosecond transient imaging” and relies on a camera equipped with a laser which emits extremely short bursts of light that can reflect off one object and then off a second before being reflecting back to the first and finally being captured by the camera. Dedicated algorithms process this information to reconstruct the hidden scene.

The research is being led by Professor Ramesh Raskar, Head of the MIT Media Lab’s Camera Culture Group, in collaboration with the Bawendi Lab in the Department of Chemistry. This ability to image hidden scenes, which the group calls “femto-photography”, involves the use of femtosecond (fs) laser illumination, picosecond-accurate, single-pixel detectors and mathematical inversion techniques. By emitting short laser pulses and analysing multi-bounce reflections it is possible to reveal hidden geometry and as light travels at ∼1 foot/nanosecond, sampling the reflected light at picosecond resolution allows the estimation of shapes with centimetre accuracy. Although the laser and high-speed cameras fit on a small optical bench, they need to be carefully calibrated before triggering. Work is at an early stage and many issues need to be resolved before practical systems emerge. These include developing a novel theoretical framework which is required to exploit fully the information contained in multi-path reflection, as no prior theory exists which models time-delayed light propagation. Moreover, constructing safe, portable device using highly experimental optics will push modern photonics technology to its limits. Nevertheless, parallel research into solid-state fs lasers will greatly simplify the illumination issue and detectors with picosecond resolution are now available for under $100. Building an array of such pixels is non-trivial but certainly feasible, although an alternative could be to use streak cameras which have a time resolution in the order of 100 fs.

Potential applications are almost limitless but some of the more obvious include search and rescue, where the technology could be used to map collapsed buildings prior to searching for survivors; collision avoidance and advanced navigation systems for cars; in machine vision and robotic guidance; and in a host of military uses. In the medical context, it could allow endoscopes to view around obstacles in the human body. In a recent paper, the researchers stated “The ability to compute the geometry of hidden elements, unobservable by both the camera and illumination source, will create a range of new computer vision opportunities.” This may well prove to be one of the greatest scientific understatements of recent times.

Rob Bogue

Further information

MIT Media Lab, available at: http://cameraculture.media.mit.edu/femtotransientimaging

MIT Femto-Photography, available at: http://web.media.mit.edu/∼raskar/femto/

Professor Ramesh Raskar; e-mail: raskar@MIT.edu