Abstract:Accurate particle detection is a common challenge in particle field characterization with digital holography, especially for gel secondary breakup with dense complex particles and filaments of multi-scale and strong background noises. This study proposes a deep learning method called Mo-U-net which is adapted from the combination of U-net and Mobilenetv2, and demostrates its application to segment the dense filament-droplet field of gel drop. Specially, a pruning method is applied on the Mo-U-net, which cuts off about two-thirds of its deep layers to save its training time while remaining a high segmentation accuracy. The performances of the segmentation are quantitatively evaluated by three indices, the positive intersection over union (PIOU), the average square symmetric boundary distance (ASBD) and the diameter-based prediction statistics (DBPS). The experimental results show that the area prediction accuracy (PIOU) of Mo-U-net reaches 83.3 % , which is about 5 % higher than that of adaptive-threshold method (ATM). The boundary prediction error (ASBD) of Mo-U-net is only about one pixel-wise length, which is one third of that of ATM. And Mo-U-net also shares a coherent size distribution (DBPS) prediction of droplet diameters with the reality. These results demonstrate the high accuracy of Mo-U-net in dense filament-droplet field recognition and its capability of providing accurate statistical data in a variety of holographic particle diagnostics. Public model address: https://github.com/Wu-Tong-Hearted/Recognition-of-multiscale-dense-gel-filament-droplet-field-in-digital-holography-with-Mo-U-net .

Short U-net model with average pooling based on in-line digital holography for simultaneous restoration of multiple particles

Characterization Method for Particle Extraction from Raw-Reconstructed Images Using U-Net

Fast non-iterative blind restoration of hyperspectral images with PSFs

B-U-net: Holographic image segmentation of multi-scale dense particle field with noisy training dataset

Recognition of Multiscale Dense Gel Filament-Droplet Field in Digital Holography With Mo-U-Net

UDC-UNet: Under-Display Camera Image Restoration via U-Shape Dynamic Network

Dense-U-net: Dense Encoder–decoder Network for Holographic Imaging of 3D Particle Fields

Holographic 3D Particle Reconstruction Using a One-Stage Network

MixNet: Efficient Global Modeling for Ultra-High-Definition Image Restoration

Dense U-net for super-resolution with shuffle pooling layer

Ultra-High-Definition Image Dehazing via Multi-Guided Bilateral Learning

Research on 3D Particle Field Reconstruction Method Based on Deep Learning

High-speed Three-Dimensional Digital Holographic Reconstruction of Particles Based on a YOLO Network

Unfocused Images Removal of Z-Axis Overlapping Mie Scattering Particles by Using Three-Dimensional Nonlinear Diffusion Based on Digital Holography.

Removal of Defocused Images Using Three-Dimensional Nonlinear Diffusion Based on Digital Holography

Accurate Detection of Small Particles in Digital Holography Using Fully Convolutional Networks.

Efficient Haze Removal from a Single Image Using a DCP-Based Lightweight U-Net Neural Network Model

A particle shape extraction and evaluation method using a deep convolutional neural network and digital image processing

Restoration and enhancement on low exposure raw images by joint demosaicing and denoising

SPH-Net: Hyperspectral Image Super-Resolution via Smoothed Particle Hydrodynamics Modeling

DU-Net: A new double U-shaped network for single image dehazing