April 2024
Spotlight Summary by Prasanna Rangarajan
Single-shot non-line-of-sight imaging based on chromato-axial differential correlography
The core elements of an imaging system, which include an illumination source and a lens assembly, have remained virtually unchanged for centuries. The design of the lens assembly is optimized for Line-of-Sight operation, while guaranteeing a one-to-one correspondence between points in the physical world and their geometric projection onto the recording medium. The notion pre-supposes that light is transported along uninterrupted ray paths originating at the illumination source, bouncing off an object and terminating at the image sensor. The emerging field of Non Line-of-Sight (NLoS) imaging challenges the above notion, by attempting to resolve spatial detail on objects that are beyond the line-of-sight and hidden from view. Such ability enables unique applications ranging from hazard avoidance in autonomous vehicles to inter-planetary exploration of caves using rovers.
The work by Liu et al. advances the state of the art in NLoS imaging by rapidly reconstituting a high-resolution image of the obscured scene from a single camera image. The task is accomplished using inexpensive off-the-shelf hardware such as RGB lasers and a color image sensor. RGB laser light scattered by an intermediary surface such as a wall is used to indirectly illuminate the hidden objects and relay the multi-spectral speckle scattered by the hidden objects towards a camera observing the intermediary surface. Previous attempts at speckle based NLoS Imaging have largely restricted attention to a single wavelength and relied on the speckle diversity afforded by scanning a laser spot on the intermediary wall to assemble a high-resolution image of the obscured objects. The work by Liu et al. dispenses with the need for scanning the laser spot by relying instead on spectral diversity of the speckle patterns produced by a multi-spectral laser source. The resulting NLoS imaging scheme affords the unprecedented ability to recover high-resolution videos of obscured objects at frame rates commensurate with Line-of-Sight imaging techniques. Other improvements over prior art include resilience to mechanical vibrations of the intermediary wall and insensitivity to spatial variation in the spectral reflectance of the intermediary wall.
The technical advances reported by Lingfeng Liu and coauthors will help accelerate the transition of NLoS imaging techniques from the lab to practical deployment in autonomous vehicles and rovers.
You must log in to add comments.
The work by Liu et al. advances the state of the art in NLoS imaging by rapidly reconstituting a high-resolution image of the obscured scene from a single camera image. The task is accomplished using inexpensive off-the-shelf hardware such as RGB lasers and a color image sensor. RGB laser light scattered by an intermediary surface such as a wall is used to indirectly illuminate the hidden objects and relay the multi-spectral speckle scattered by the hidden objects towards a camera observing the intermediary surface. Previous attempts at speckle based NLoS Imaging have largely restricted attention to a single wavelength and relied on the speckle diversity afforded by scanning a laser spot on the intermediary wall to assemble a high-resolution image of the obscured objects. The work by Liu et al. dispenses with the need for scanning the laser spot by relying instead on spectral diversity of the speckle patterns produced by a multi-spectral laser source. The resulting NLoS imaging scheme affords the unprecedented ability to recover high-resolution videos of obscured objects at frame rates commensurate with Line-of-Sight imaging techniques. Other improvements over prior art include resilience to mechanical vibrations of the intermediary wall and insensitivity to spatial variation in the spectral reflectance of the intermediary wall.
The technical advances reported by Lingfeng Liu and coauthors will help accelerate the transition of NLoS imaging techniques from the lab to practical deployment in autonomous vehicles and rovers.
Add Comment
You must log in to add comments.
Article Information
Single-shot non-line-of-sight imaging based on chromato-axial differential correlography
Lingfeng Liu, Shuo Zhu, Wenjun Zhang, Lianfa Bai, Enlai Guo, and Jing Han
Photon. Res. 12(1) 106-114 (2024) View: Abstract | HTML | PDF