Corpus ID: 231709519

Real-time Non-line-of-sight Imaging with Two-step Deep Remapping

  title={Real-time Non-line-of-sight Imaging with Two-step Deep Remapping},
  author={Dayu Zhu and W. Cai},
Conventional imaging only records the photons directly sent from the object to the detector, while non-line-of-sight (NLOS) imaging takes the indirect light into account. To explore the NLOS surroundings, most NLOS solutions employ a transient scanning process, followed by a back-projection based algorithm to reconstruct the NLOS scenes. However, the transient detection requires sophisticated apparatus, with long scanning time and low robustness to ambient environment, and the reconstruction… Expand

Figures from this paper


Deep Non-Line-of-Sight Reconstruction
This paper devise a tailored autoencoder architecture, trained end-to-end, that maps transient images directly to a depth-map representation and demonstrates that the feed-forward network employed, even if trained solely on synthetic data, is able to obtain results competitive with previous, model-based optimization methods, while being orders of magnitude faster. Expand
Real-time Non-line-of-Sight imaging of dynamic scenes
It is shown that SPAD (Single-Photon Avalanche Diode) array detectors with a total of just 28 pixels combined with a specifically extended Phasor Field reconstruction algorithm can reconstruct live real-time videos of non-retro-reflective NLOS scenes. Expand
Steady-State Non-Line-Of-Sight Imaging
This work departs from temporal probing and demonstrates steady-state NLOS imaging using conventional intensity sensors and continuous illumination, and proposes a trainable architecture which learns to map diffuse indirect reflections to scene reflectance using only synthetic training data. Expand
Confocal non-line-of-sight imaging based on the light-cone transform
The resolution bounds of NLOS imaging are quantified, its potential for real-time tracking is demonstrated, and efficient algorithms that incorporate image priors and a physically accurate noise model are derived. Expand
Deep-inverse correlography: towards real-time high-resolution non-line-of-sight imaging
This work uses spectral estimation theory to derive a noise model for NLoS correlography, and develops a speckle correlation-based technique for recovering occluded objects from indirect reflections, and trains a deep convolutional neural network to solve the noisy phase retrieval problem associated with correlography. Expand
Fast Back-Projection for Non-Line of Sight Reconstruction
This work proposes a new back-projection technique for NLOS reconstruction, up to a thousand times faster than previous work, with almost no quality loss, and demonstrates the efficiency and quality of the technique compared against previous methods. Expand
Convolutional Approximations to the General Non-Line-of-Sight Imaging Operator
This work introduces a computationally tractable framework for solving the ellipsoidal tomography problem by using efficient deconvolution procedures to solve a linear least-squares problem involving the Gram operator. Expand
The Geometry of First-Returning Photons for Non-Line-of-Sight Imaging
This paper shows that the times of flight of first-returning photons are dependent only on the geometry of the NLOS scene and each observation is almost always generated from a singleNLOS scene point, and derives a space carving algorithm for NLOS scenes by assuming local planarity. Expand
Exploiting Occlusion in Non-Line-of-Sight Active Imaging
It is demonstrated that the presence of occluders in the hidden scene can obviate the need for collecting time-resolved measurements, and an accompanying analysis for such systems and their generalizations is developed. Expand
Revealing hidden scenes by photon-efficient occlusion-based opportunistic active imaging.
This work employs a computational imaging technique that opportunistically exploits the presence of occluding objects, which obstruct probe-light propagation in the hidden scene, to undo the mixing and greatly improve scene recovery and represents an instance of a rich and promising new imaging modality. Expand