Single-shot super-resolved fringe projection profilometry (SSSR-FPP): 100,000 frames-per-second 3D imaging with deep...

Area of Science:

• Optical sensing
• 3D imaging
• Dynamic processes

Background:

• Ultrafast 3D imaging is crucial for studying transient events in mechanics, physics, and biology.
• Conventional fringe projection profilometry (FPP) is limited to kHz speeds due to hardware and multi-pattern requirements.

Purpose of the Study:

• To develop a novel learning-based technique for ultrafast 3D imaging.
• To overcome the speed limitations of traditional FPP methods.

Main Methods:

• Introduced single-shot super-resolved FPP (SSSR-FPP), a deep learning-based approach.
• Utilized a trained deep neural network to decipher high-resolution phase and fringe orders from single, low-SNR, low-resolution fringe patterns.
• Leveraged reduced imaging windows of high-speed cameras and deep learning for spatial resolution "regeneration".

Main Results:

• Achieved ultrafast 3D imaging at 100,000 Hz.
• Successfully captured 3D videography of dynamic events like rotating turbofan blades and exploding building blocks.
• Demonstrated high spatio-temporal resolution surpassing conventional methods.

Conclusions:

• SSSR-FPP represents a significant advancement in 3D optical sensing.
• Enables new insights into a wide range of dynamic processes previously difficult to capture.
• Offers a powerful tool for scientific disciplines requiring high-speed 3D imaging.