Sub-picosecond photon-efficient 3D imaging using single-photon sensors.

Area of Science:

• Optics and Photonics
• Computer Vision
• Robotics

Background:

• Active 3D imaging systems are crucial for applications in biology, remote sensing, and robotics.
• Single-photon avalanche diodes (SPADs) offer high speed, timing accuracy, and sensitivity but suffer from pileup distortions.
• Pileup fundamentally limits the precision of SPAD-based 3D imaging systems.

Purpose of the Study:

• To develop a probabilistic image formation model that accurately accounts for pileup distortions in SPADs.
• To devise inverse methods for robustly estimating scene depth and reflectance from photon counts using the pileup model.
• To enable high-precision, photon-efficient 3D imaging in challenging, real-world scenarios.

Main Methods:

• Development of a probabilistic image formation model specifically designed to handle pileup in SPAD detectors.
• Implementation of inverse methods incorporating statistical priors for accurate scene depth and reflectance estimation.
• Validation of the proposed algorithm in practical 3D imaging scenarios with wide variations in photon counts.

Main Results:

• Demonstrated an order of magnitude improvement in timing accuracy compared to existing state-of-the-art methods.
• Achieved sub-picosecond accuracy in 3D imaging, a first for practical, photon-efficient applications.
• Successfully enabled precise 3D imaging in scenarios with highly variable photon counts, overcoming pileup limitations.

Conclusions:

• The developed probabilistic model and inverse methods effectively mitigate pileup distortions in SPAD-based 3D imaging.
• This work significantly advances the state-of-the-art in 3D imaging by enabling unprecedented timing accuracy and efficiency.
• The proposed approach paves the way for more robust and precise 3D sensing in diverse scientific and industrial fields.