Researchers at Massachusetts Institute of Technology, Harvard, and Rice University have devised a system that can produce recognizable 3-D images of objects outside of a camera’s line of sight. Their findings are described in this week’s issue of the journal Nature Communications (paid subscription required).
The interdisciplinary team of engineers, mathematician, and chemist, led by Ramesh Raskar, a professor in MIT’s Media Lab, developed the system based on a technology called femtosecond lasers. These lasers emit extremely short bursts of light measured in quadrillionths of a second, but can also reflect off surfaces not usually considered reflective, such as walls, doors, and floors.
Light bursts from a femtosecond laser, sent from outside a doorway into a room, for example, could bounce off the walls and floors then re-emerge, to be captured by a detector taking measurements in picoseconds — one trillion frames per second. The measurements then indicate the distance the light bursts have traveled by measuring the time it takes them to reach the detector.
The system performs this process from several angles, bouncing light bursts off different targets and surfaces in the room, taking time and distance measurements for each burst. The system can then assemble the measurements into a composite image based on the geometry of the room.
Femtosecond lasers had been used previously to produce high-speed images in the lab of biochemical processes, which led to the use of equipment in the lab of MIT chemist Moungi Bawendi, a co-author of the paper. Andreas Velten, a postdoc at the time in Raskar’s group, conducted the experiments. Velten, Raskar, with Thomas Willwacher, a mathematics postdoc at Harvard University, and Ashok Veeraraghavan, a professor of electrical engineering and computer science at Rice University developed the algorithms for processing the images.
The images from the experiments show a wooden figurine (pictured at top) and foam cutouts outside the camera’s field of vision. The 3-D images are blurry, but still recognizable. The discovery, say the researchers, can lead to emergency response technologies in hazardous environments and navigation systems for vehicles negotiating blind turns. The technology could also be used to help make endoscopic medical devices less invasive for images taken inside the human body.
The following video, courtesy of the MIT Media Lab, tells more about the femto-photography process.
* * *