In the report ‘Confocal non-line-of-sight imaging based on the light-cone transform’, published yesterday in Nature, the team describes how they set up a laser next to a photon detector which can record a single particle of light. The laser then shoots light, which is invisible to the human eye, off a wall and onto objects around the corner, which then bounces back off the wall and to the detector.
This scan can take between two minutes and one hour, based on lighting and the reflectivity of the object around the corner, and once completed an algorithm produces a sharp image of the area. The researchers think that the algorithm could be tweaked to produce an image nearly immediately once the scan is complete.
“A substantial challenge in non-line-of-sight imaging is figuring out an efficient way to recover the 3D structure of the hidden object from the noisy measurements,” said David Lindell, co-author of the paper and graduate student in the Stanford Computational Imaging Lab.
“I think the big impact of this method is how computationally efficient it is.”
The technology could be used to help autonomous vehicles navigate. Many cars are currently equipped with LIDAR systems, means of surveying the immediate environment by illuminating areas with pulsed laser light and measuring the reflected pulses with a sensor. While many LIDAR systems intentionally ignore scattered light particles so as to focus on specific objects rather than building up a profile of the wider environment, Matthew O’Toole, lead co-author of the paper, believes LIDAR hardware can be effectively adapted generate images of larger areas.
“We believe the computation algorithm is already ready for LIDAR systems,” said O’Toole, a postdoctoral scholar at the Computational Imaging Lab.
“The key question is if the current hardware of LIDAR systems supports this type of imaging.”
“It sounds like magic but the idea of non-line-of-sight imaging is actually feasible,” added Gordon Wetzstein, senior author of the paper and assistant professor of electrical engineering.
“This is a big step forward for our field that will hopefully benefit all of us. In the future, we want to make it even more practical ‘in the wild.’”