Abstract:Compressed ultrafast photography (CUP) is the fastest single-shot passive ultrafast optical imaging technique, which has shown to be a powerful tool in recording self-luminous or non-repeatable ultrafast phenomena. However, the low fidelity of image reconstruction based on the conventional augmented-Lagrangian (AL) and two-step iterative shrinkage/thresholding (TwIST) algorithms greatly prevents practical applications of CUP, especially for those ultrafast phenomena that need high spatial resolution. Here, we develop a novel AL and deep-learning (DL) hybrid (i.e., <math display="inline"> <mrow> <mi>AL</mi> <mo>+</mo> <mi>DL</mi> </mrow> </math> ) algorithm to realize high-fidelity image reconstruction for CUP. The <math display="inline"> <mrow> <mi>AL</mi> <mo>+</mo> <mi>DL</mi> </mrow> </math> algorithm not only optimizes the sparse domain and relevant iteration parameters via learning the dataset but also simplifies the mathematical architecture, so it greatly improves the image reconstruction accuracy. Our theoretical simulation and experimental results validate the superior performance of the <math display="inline"> <mrow> <mi>AL</mi> <mo>+</mo> <mi>DL</mi> </mrow> </math> algorithm in image fidelity over conventional AL and TwIST algorithms, where the peak signal-to-noise ratio and structural similarity index can be increased at least by 4 dB (9 dB) and 0.1 (0.05) for a complex (simple) dynamic scene, respectively. This study can promote the applications of CUP in related fields, and it will also enable a new strategy for recovering high-dimensional signals from low-dimensional detection.

Deep Learning Reconstruction of Ultra-Short Pulses

Engineering spectro-temporal light states with physics-trained deep learning

Neural-network-powered pulse reconstruction from one-dimensional interferometric cross-correlation traces

A novel dynamic scheme to reconstruct sub-50 atto-second pulses

Deep neural networks for high harmonic spectroscopy in solids

Femtosecond pulse parameter estimation from photoelectron momenta using machine learning

Fast reconstruction of laser beam near-field and focal spot profiles using deep neural network and numerical propagation

Physics Constrained Unsupervised Deep Learning for Rapid, High Resolution Scanning Coherent Diffraction Reconstruction

Artificial intelligence for online characterization of ultrashort X-ray free-electron laser pulses

Discrete Illumination‐Based Compressed Ultrafast Photography for High‐Fidelity Dynamic Imaging (Adv. Sci. 41/2024)

Deep-sub-cycle ultrafast optical pulses

Virtual Pulse Reconstruction Diagnostic for Single-Shot Measurement of Free Electron Laser Radiation Power

Attosecond electron pulse trains and quantum state reconstruction in ultrafast transmission electron microscopy

Efficient prediction of attosecond two-colour pulses from an X-ray free-electron laser with machine learning

Diagnosis of ultrafast ultraintense laser pulse characteristics by machine-learning-assisted electron spin

Terahertz Pulse Shaping Using Diffractive Surfaces

Single-Shot, Spatio-Temporal Metrology of Relativistic Plasma Optics

High-fidelity image reconstruction for compressed ultrafast photography via an augmented-Lagrangian and deep-learning hybrid algorithm

Low Latency Computing for Time Stretch Instruments

Digital Holography for Real-time Detection Based on Femtosecond Laser

Shot-Noise Limited Nonlinear Optical Imaging Excited With GHz Femtosecond Pulses and Denoised by Deep-Learning