Start from the peer-reviewed survey, then use the continuously updated companion survey and interactive research map to follow new methods, modalities and milestones through 2026.
Ruixu Geng, Yang Hu, and Yan Chen. APSIPA Transactions on Signal and Information Processing, 2022.
A continuously maintained companion to the 2022 paper, integrating newly verified optical, passive, learned, LiDAR, radar/RF/mmWave, acoustic, THz and emerging NLOS directions. This extension is a living project document rather than a separately peer-reviewed publication.
Each paper is a node. Color encodes publication year, node size reflects how strongly a paper sits inside the co-author network, and links connect papers that share authors. Search by paper, researcher, year or research family.
Use the trends as a guide to the repository's coverage: when activity accelerated, which research families expanded, and which researchers repeatedly appear across the curated literature.
Publication-year distribution within this curated NLOS corpus.
Active, passive, learning, RF/mmWave, acoustic and related directions.
Start with the field-defining Nature papers, then follow the experimental, algorithmic, learning and modality-expansion trajectory.
Newly integrated historical and transient-imaging links: consumer-camera periscopy (SPIE 2019) and door-frame 3D passive imaging (IEEE RAPID 2022) precede the mature TERI system; temporal-decoupling NLOS super-resolution (Optics and Lasers in Engineering 2026) extends the transient reconstruction branch. PDF rebuild pending.
Coherent synthetic-wavelength milestone: Willomitzer et al. (Nature Communications 2021) recover hidden geometry with high resolution and wide field of view using optical-interference-derived synthetic wavelengths, complementing transient ToF and phasor-field systems.
Optical hidden-physiology frontier: Chen et al. (Optica 13(9), 2026) extend indirect optical NLOS sensing to hidden vital signs; this is physiological sensing rather than dense 3D scene reconstruction.
Passive computational-periscopy and thermal precursors: Doorway Camera (ICCP 2022) and occlusion-aided 3D NLOS (RAPID 2024) explore aperture coding without transient capture; Hashemi et al. (ICCP 2023) exploit thermal parallax to recover weak hidden-person signals.
Active–passive color/geometry fusion: Shi et al. (Photonics Research 2026, accepted 14 October) jointly recover hidden 3D geometry and color by decomposing active and passive measurements into complementary amplitude and color information; available as an early-posting article pending final issue assignment.
Passive learned reconstruction and display-side-channel branch: Zhou et al. (Photonics 2026) recover bounded hidden-screen structure from a single diffuse-wall RGB image using survivability-guided feature fusion. The reported results are restricted to controlled imaging conditions.
Passive aperture and multi-object reconstruction lineage: Shi et al. (Optics Express 2024) extends computational-periscopy-like geometry from edge occluders to large transmissive windows, using depth separation to reconstruct multiple overlapping hidden objects with an ordinary camera.
Polarization and multispectral passive NLOS: PI-NLOS (Optics Express 2023) fuses polarized LWIR intensity and degree-of-polarization cues; multispectral deep fusion (Science China Information Sciences 2025) combines visible, SWIR and LWIR observations for hidden-person reconstruction.
Newly verified reconstruction and acquisition directions: TriNLOS (ECCV 2026) uses triplane neural volumes after learned LCT; NLOSdiffuser (ICME 2025) combines steady-state inversion with diffusion priors; and gradient-gated SPAD (ESSERC 2025) advances transient acquisition hardware.
Earlier acquisition and hardware lineage: foveated NLOS (COSI 2020) proposed transient-plus-correlography capture; picosecond-gated NLOS (Optics Express 2021) demonstrated photon-efficient hidden-target and vibration sensing; fractal SNSPD NLOS (Optics Express 2023) demonstrated 1560/1997 nm reconstruction. The first is a conference digest; the latter two are journal papers.
Fast coherent NLOS acquisition: Huang et al. (PRL 2024) established comb-calibrated FMCW hidden-scene imaging and vibrometry; Qian et al. (SPIE Photonics Europe 2026) demonstrated predistorted near-infrared FMCW LiDAR with approximately 0.2 s acquisition per NLOS frame. The latter is a conference demonstration, not a journal result.
Adjacent NLOS sensing and localization (not full hidden-scene imaging): PRISM (MobiHoc 2025) uses RIS frequency shifts for mmWave target localization; Mao et al. (IWS 2026) recover hidden vital signs; Tian et al. (2024) use equivalent-source time reversal for hidden acoustic-source localization; and Sroka et al. (JASA 2026) incorporate rigid-boundary acoustics. Earlier acoustic source-localization milestones (distinct from hidden-scene imaging) include Mak & Furukawa (AIM 2009), Singh et al. (JASA 2012), Kitić et al. (ICASSP 2014), Takami et al. (IROS 2016), and An et al. (ICRA 2019). Chakraborty et al. (SPAWC 2025) and King et al. (Applied Acoustics 2021) expand radar and acoustic NLOS localization. Explore the full RF and acoustic branches in the Paper Explorer.
Verified RF/multipath developments: Chen et al. (IEEE TRS 2025) jointly localize LOS and NLOS targets under clutter; Zhang et al. (IEEE TWC 2026) recover double-bounce radio-SLAM landmarks; and Tong et al. (IEEE TWC 2026) couple distributed-MIMO NLOS imaging with over-the-air synchronization. These are RF mapping/localization advances rather than dense hidden-object reconstruction.
Search the complete deduplicated corpus without rendering hundreds of large cards at once.
If you find this website useful in your research, please consider citing our survey paper:
@article{geng2022recent,
title = {Recent Advances on Non-Line-of-Sight Imaging: Conventional Physical Models, Deep Learning, and New Scenes},
author = {Geng, Ruixu and Hu, Yang and Chen, Yan},
journal = {APSIPA Transactions on Signal and Information Processing},
volume = {11},
number = {1},
year = {2022},
doi = {10.1561/116.00000019}
}