Non-line-of-sight (NLOS) imaging is a cutting-edge technology that enables the visualization of objects hidden from direct view, such as those situated around corners or behind obstacles. This capability has significant implications across various fields, from security and surveillance to autonomous vehicles and medical imaging. By harnessing NLOS imaging, it becomes feasible to detect and identify objects or individuals that would otherwise remain obscured from view, thus enhancing situational awareness and decision-making processes.
The principle behind NLOS imaging involves the detection of faint light reflections that have scattered multiple times before reaching the imaging sensor. This scattered light carries information about the hidden objects' surfaces and shapes, allowing their reconstruction even when they are not directly visible. However, capturing these faint reflections presents a considerable technical challenge, as traditional imaging sensors may not be sensitive enough to detect such weak signals reliably.
To address this challenge, specialized image sensors have been developed, but to date, they are only capable of operating in the visible and near-infrared spectrum of light. Unfortunately, background irradiance from the sun can wreak havoc on the algorithms that process this data to reconstruct a hidden scene. By instead sensing longer wavelengths of infrared light, this interference could be greatly diminished, enabling more accurate NLOS sensing capabilities for important applications like self-driving cars.
The problem is that image sensors with the necessary sensitivity at these wavelengths do not exist. Or rather they did not exist until recently. A team at Tianjin University has developed an image sensor that is so sensitive in the infrared range that it can even detect single photons. The technology might not be entirely practical to use outside of laboratory conditions just yet — it must be supercooled — but it is very effective. The device was shown to be about three times more efficient than existing single-photon detectors at sensing infrared light.
The sensor contains a superconducting nanowire composed of niobium titanium nitride. When cooled to approximately -520 degrees Fahrenheit, the wire acquires superconductive properties. When in this state, even a single photon can disturb it. That, in turn, generates an electrical pulse that signals the presence of a photon within its range of detectable wavelengths. To ensure that photons of any polarization can be captured, the nanowire was bent into a specific fractal pattern.
Designing the sensor was a crucial piece of the puzzle, but not enough to see around corners with infrared light. To accomplish that, the researchers had to develop a pair of algorithms to process the sensor measurements. The first reconstructs NLOS images from the light received by the sensor. The second algorithm denoises this initial result. This added step results in the production of much clearer final images.
Encouraged by the successes they have seen thus far, the researchers are planning to improve their technology in the future. They intend to investigate ways that additional wavelengths of interest can be sensed using similar techniques, hoping to unlock even more potential applications. The team is also exploring the possibility of creating an array of single-photon sensors to increase the collection efficiency of the system, and also to reduce scanning times. With enhancements like these, many use cases could ultimately benefit from new NLOS technologies.