Particle positioning and characterization using wavefront curvature with a convolutional neural network in digital...

Area of Science:

• Optical Physics
• Computational Imaging
• Particle Metrology

Background:

• Accurate simultaneous determination of spherical particle parameters (axial position, radius, refractive index) is challenging in digital holography (DH).
• Low-numerical-aperture (NA) optical setups, common in applications like fluid dynamics and aerosol characterization, provide a wide field of view but yield fewer interference fringes, complicating conventional analysis.
• This limitation hinders precise parameter estimation crucial for many scientific and industrial applications.

Purpose of the Study:

• To develop and validate a novel method for accurate and simultaneous determination of spherical particle parameters in digital holography.
• To overcome the limitations of conventional analysis, particularly in low-NA optical systems.
• To enhance the capabilities of DH for particle characterization across various NA configurations.

Main Methods:

• A one-dimensional convolutional neural network (1D-CNN) was developed and trained.
• The network was trained using wavefront curvature profiles along the optical axis.
• The method was tested on both low-NA (0.02) and high-NA systems, with experimental validation.

Main Results:

• The proposed method accurately and simultaneously determines particle radius, axial position, and refractive index.
• In a low-NA setup, the method achieved high accuracy: 0.3% for radius, 2.0% for axial position, and 7.0% for refractive index.
• Performance significantly outperformed conventional holographic interference pattern analysis methods.

Conclusions:

• The 1D-CNN approach effectively addresses the challenges in DH particle parameter estimation, especially for low-NA systems.
• This method enhances DH capabilities for particle characterization in diverse applications requiring any NA.
• The validated technique offers a robust solution for precise, simultaneous multi-parameter determination of spherical particles.