Seeing Around Corners with Edge-Resolved Transient Imaging (2002.07118v1)

Abstract: Non-line-of-sight (NLOS) imaging is a rapidly growing field seeking to form images of objects outside the field of view, with potential applications in search and rescue, reconnaissance, and even medical imaging. The critical challenge of NLOS imaging is that diffuse reflections scatter light in all directions, resulting in weak signals and a loss of directional information. To address this problem, we propose a method for seeing around corners that derives angular resolution from vertical edges and longitudinal resolution from the temporal response to a pulsed light source. We introduce an acquisition strategy, scene response model, and reconstruction algorithm that enable the formation of 2.5-dimensional representations -- a plan view plus heights -- and a 180$^{\circ}$ field of view (FOV) for large-scale scenes. Our experiments demonstrate accurate reconstructions of hidden rooms up to 3 meters in each dimension.

• The paper introduces Edge-Resolved Transient Imaging (ERTI), a novel non-line-of-sight method that uses edge occlusions and transient light signals to reconstruct occluded 2.5D scenes with a 180-degree field of view.
• ERTI utilizes a pulsed laser scanned along an arc, a single-photon detector capturing time-resolved signals, and Bayesian inference for robust reconstruction of hidden scene structures like distances, heights, and orientations.
• Experimental validation shows ERTI accurately determines object dimensions within tens of centimeters even in noisy conditions, with potential applications in rescue, reconnaissance, and healthcare imaging.

Seeing Around Corners with Edge-Resolved Transient Imaging

This paper introduces a method for non-line-of-sight (NLOS) imaging called Edge-Resolved Transient Imaging (ERTI), offering potential advancements in imaging scenarios where the objects under paper are outside the direct line of sight. The significance of this work lies in its ability to derive angular and longitudinal resolution in NLOS scenarios through a novel combination of active and passive methods, facilitating the reconstruction of complex scenes with a 180-degree field of view. The approach presented amalgamates strategic acquisitions with sophisticated reconstruction algorithms, supporting the generation of 2.5-dimensional representations of sizable, occluded scenes.

Methodology and Light Transport Model

ERTI leverages the common occurrence of vertical edges, such as door frames, to gain directional insight into occluded areas. The researchers exploit the geometry formed by scanning a pulsed laser along an arc and employing a single-photon-sensitive detector to capture reflected light signals, integrating time-resolved acquisition with edge occlusions. Differences in successive photon detection histograms isolate regions within hidden scenes, mitigating the directional uncertainty typically associated with scatter-induced reflections in conventional NLOS imaging methodologies. The light transport model, tailored for this transient imaging setup, enables computationally efficient descriptions of planar facets, characterized by their spatial and reflective properties.

Reconstruction Algorithm

The reconstruction algorithm functions via Bayesian inference, employing a Markov chain Monte Carlo method to derive the facet configurations from empirical data. This algorithm dynamically adjusts the number and arrangement of facets, offering robust recovery of hidden scene structures despite the intrinsic signal noise and complexity inherent in real-world data. By targeting spatial dependencies and regularities within the hidden scenes, the technique consistently produces accurate reconstructions of room configurations and background elements, revealing component distances, heights, orientations, and albedos with marked precision.

Experimental Validation

The research emphasizes empirical validation through indoor scene experiments involving mannequins and architectural elements like staircases. The results elucidate the approach's efficacy in determining object dimensions to within tens of centimeters, even amidst high ambient noise conditions. The efficiency of the proposed method is evident in its execution time being significantly shorter than conventional acquisition times, suggesting foreseeable improvements in NLOS imaging systems.

Discussion and Future Directions

The paper proposes enhancements to existing systems through increased laser power and multi-wavelength setups for greater eye safety and efficiency. There is potential for exploiting a broader array of occluder configurations and adjusting acquisition strategies to improve resolution and capture dynamic scenes effectively. The manner in which ERTI integrates geometric constraints with transient light-signals suggests further exploration into adaptive, intelligently designed imaging networks capable of recovering even more complex occluded environments. This work lays the groundwork for the potential advancement of applications in various fields, including rescue operations, reconnaissance, and healthcare imaging contexts, modulating traditional visual paradigms with sophisticated computational insight.