Deep Vision: challenging the limits of imaging through obscurants by light-field photon counting
Abstract
Imaging through obscurants is one of the most desirable and exciting abilities for many applications at different scales, ranging from assisted driving and target identification in challenging scenarios (e.g. fog, smoke) to non-invasive imaging deep inside the human body. However, photons at visible and near-infrared wavelengths propagating within obscurants suffer from uncontrolled scattering. This modifies the optical wavefront severely and dramatically attenuates light intensity upon propagation. The direct consequenceis that detecting, identifying, and imaging targets through or inside the obscurant remains an outstanding challenge. However, during the last decade the advancement in computational imaging techniques and ultrasensitive imaging technologies has shone new light into the field. In this proposal we will explore in detail the problem of imaging over 50-100 transport mean free paths (TMFPs). Our work will be based on a series of recent innovations from our group: information theory analysis of spatio-temporal data if light propagating in scattering media; novel AI approaches for decoding diffuse photons into images; novel proposals for next generation single-photon time-of-flight imaging that will allow to utilise unexplored regions of the plenotopic imaging function. By bridging these novel computational and hardware approaches, we expect to push the boundaries of current imaging capability with the explicit goalof breaking the 100 TMFP barrier.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jun 09, 2021
- Source ID
- N629092112027
Entities
People
- Daniele Faccio
Organizations
- Office of Naval Research
- United States Navy
- University of Glasgow