Digital holographic particle volume reconstruction using a deep neural network.

Area of Science:

• Optics and Photonics
• Computational Imaging
• Particle Metrology

Background:

• Digital holographic volume reconstruction typically involves multiple diffraction calculations.
• Conventional methods suffer from limited axial resolution and are time-consuming.
• Focus metrics are used to detect particle properties, but have limitations.

Purpose of the Study:

• To propose a novel method for direct particle volume reconstruction from in-line holograms.
• To develop a deep neural network (DNN) capable of simultaneous detection of particle properties.
• To overcome the limitations of conventional holographic reconstruction techniques.

Main Methods:

• A deep neural network (DNN) was designed for direct reconstruction from in-line holograms.
• The DNN was trained to simultaneously detect lateral and axial positions, and particle sizes.
• Numerical simulations were conducted to evaluate the DNN's performance.

Main Results:

• The DNN achieved simultaneous detection of particle lateral and axial positions and sizes.
• Numerical investigations quantified the errors in detected positions and sizes.
• The proposed DNN method demonstrated significantly faster calculation times compared to conventional approaches.

Conclusions:

• The DNN-based approach offers a faster and potentially more accurate method for particle volume reconstruction.
• This technique overcomes the axial resolution limitations inherent in conventional optical systems.
• The study highlights the potential of deep learning in advancing holographic particle analysis.