Angular spectrum-encoded single-shot ultrafast photography.

Area of Science:

• Optics and Photonics
• High-Speed Imaging
• Machine Learning Applications

Background:

• Capturing transient events requires imaging at trillions of frames per second (Tfps).
• Existing methods like compressed sensing and time-resolved shadowgraphy are limited by bulky optics, high cost, and repeated measurements.
• There is a need for compact, cost-effective, and single-shot ultrafast imaging techniques.

Purpose of the Study:

• To introduce a novel ultrafast imaging technique called angular spectrum-encoded single-shot ultrafast photography (ASUP).
• To demonstrate ASUP's capability for high-fidelity reconstruction of ultrafast dynamics in a single exposure.
• To overcome the limitations of existing ultrafast imaging methods.

Main Methods:

• ASUP combines time-wavelength mapping with dispersion-encoded angular spectrum information.
• A multilayer dielectric thin-film photonic chip, designed using a deep Q-network (DQN) reinforcement-learning framework, performs pixel-level encoding.
• An enhanced residual convolutional neural network with Transformer blocks decodes measurements for image reconstruction.

Main Results:

• ASUP achieves an imaging rate of 0.83 Tfps, capturing six frames in a single exposure.
• The technique successfully imaged picosecond laser-induced damage and plasma dynamics in metal films.
• Reconstructed images demonstrate high fidelity, comparable to state-of-the-art methods.

Conclusions:

• ASUP offers a compact, integrated, and cost-effective solution for ultrafast photography.
• The method overcomes key limitations of current ultrafast imaging techniques.
• ASUP is a scalable solution for high-speed optical diagnostics, laser-matter interactions, and transient phenomena studies.