Being able to see inside a closed room was a skill previously reserved for superheroes. But researchers from the Stanford Computational Imaging Laboratory have expanded a technique called line-of-sight imaging so that a single point of laser light entering a room can be used to see what physical objects might be inside.
Non-line of sight imaging (NLOS for short) is by no means a new idea. It is a clever technique that has been perfected in research labs over the years to create cameras that can noticeably see around corners and generate object images that would otherwise not be in the camera’s field of view or that are blocked by a series of obstacles. Previously, the technique has taken advantage of flat surfaces such as floors or walls that are in the line of sight of both the camera and the obstructed object. A series of light pulses originating from the camera, usually from lasers, bounce off these surfaces and then bounce off the hidden object before finally returning to the camera’s sensors. The algorithms then use the information about how long it took for these reflections to return to generate an image of what the camera cannot see. The results are not high resolution, but are usually detailed enough to easily determine which object is in question.
It’s an incredibly clever technique, and it could one day be a very useful technology for devices like self-driving cars that could potentially detect potential dangers hidden in corners long before they’re visible to passengers in a vehicle, improving safety and avoiding obstacles. But current NLOS techniques have one major limitation: they rely on a large reflective surface where reflections of light from a hidden object can be measured. Trying to imagine what is inside a closed room from the outside is almost impossible, or at least it was until now.
The keyhole imaging technique, developed by researchers at Stanford University’s Computer Imaging Laboratory, is so named because all that is needed to see what’s inside a closed room is a small hole (like an eye keyhole or peephole) large enough for a laser beam to pass through, creating a single point of light on an interior wall. As in previous experiments, the laser light bounces off a wall, an object in the room, and then the wall again, with countless photons eventually being reflected through the hole and into the camera using a photon avalanche photodetector. unique to measure the time of your return.
When a hidden object in the room is static, the new keyhole imaging technique simply cannot calculate what you are seeing. But researchers have found that a moving object paired with pulses of light from a laser generates enough usable data over a long period of exposure time for an algorithm to create an image of what it is seeing. The quality of the results is even worse than with previous NLOS techniques, but it still provides enough detail to make an educated guess about the size and shape of the hidden object. A wooden mannequin ends up looking like a ghostly angel, but when combined with a properly trained image-recognition AI, determining that a human (or human-shaped object) was in the room seems very doable.
The investigation could one day provide a way for the police or military to assess the risks of entering a room before smashing the door and storming inside, using nothing more than a small crack in the wall or a space around a window. or entrance. The new technique could also provide new techniques for autonomous navigation systems to detect hidden hazards long before they become a threat in situations where previous NLOS techniques were impractical given the environment.