Category Ranking
98%
Total Visits
921
Avg Visit Duration
2 minutes
Citations
20
Article Abstract
Passive non-line-of-sight (NLOS) imaging has potential applications in autonomous driving and search and rescue, but current deep learning approaches often produce suboptimal images due to sparse and homogeneous projection features, leading to an ill-posed reconstruction process. To address this, we propose the Hyperspectral Fusion NLOS imaging technique (HFN-Net), which first leverages high-dimensional features from multiple spectra and incorporates spatial-spectral attention from a hyperspectral full-color auto-encoder. This method improves color fidelity and structural details by fully utilizing the limited information and increasing feature diversity. Additionally, we developed the Hyperspectral NLOS dataset (HS-NLOS) for training and evaluation. Experimental results show that HFN-Net offers performance improvements over traditional passive NLOS 2D imaging techniques, emphasizing the importance of multi-spectral information.
Download full-text PDF |
Source |
|---|---|
| http://dx.doi.org/10.1364/OE.532699 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
Non-line-of-sight (NLOS) imaging is an inverse problem that consists of reconstructing a hidden scene out of the direct line-of-sight given the time-resolved light scattered back by the hidden scene on a relay wall. Phasor fields transforms NLOS imaging into virtual LOS imaging by treating the relay wall as a secondary camera, which allows reconstruction of the hidden scene using a forward diffraction operator based on the Rayleigh-Sommerfeld diffraction (RSD) integral. In this work, we leverage the unitary property of the forward diffraction operator and the dual space it introduces, concepts already studied in inverse diffraction, to explain how phasor fields can be understood as an inverse diffraction method for solving the hidden object reconstruction, even though initially it might appear it is using a forward diffraction operator.
View Article and Find Full Text PDFWe develop a framework for thermal passive non-line-of-sight (NLOS) imaging based on the reconstruction of noisy scattered light fields using neural networks. Thermal NLOS imaging is undermined by the extremely low reflectivity and relatively diffuse nature of many surfaces in the long-wave-infrared (LWIR) domain. Previous approaches in thermal NLOS imaging have relied on linear methods to denoise and deblur measurements.
View Article and Find Full Text PDFNon-line-of-sight (NLOS) imaging is an emerging computational imaging technique that allows acquiring information about hidden objects from an indirect view. It has garnered significant attention due to its great potential in medical imaging, security, and autonomous driving, among many others. However, most existing NLOS techniques require active coherent illumination and expensive equipment, which prevents their practical application.
View Article and Find Full Text PDFNon-line-of-sight (NLOS) imaging is a rapidly developing technology with significant potential applications. Achieving fast and high-resolution imaging is crucial in practical scenarios. In existing NLOS imaging methods, most approaches rely on time-consuming scanning to detect scenes, which makes it challenging to achieve rapid acquisition.
View Article and Find Full Text PDFPassive non-line-of-sight (NLOS) imaging has potential applications in autonomous driving and search and rescue, but current deep learning approaches often produce suboptimal images due to sparse and homogeneous projection features, leading to an ill-posed reconstruction process. To address this, we propose the Hyperspectral Fusion NLOS imaging technique (HFN-Net), which first leverages high-dimensional features from multiple spectra and incorporates spatial-spectral attention from a hyperspectral full-color auto-encoder. This method improves color fidelity and structural details by fully utilizing the limited information and increasing feature diversity.
View Article and Find Full Text PDF