Real-Time Structural Illumination with Hyperspectral Images: A Tunable Projection-Capture Synchronizer for...

Area of Science:

• Optics and Photonics
• Computer Vision
• Embedded Systems Engineering

Background:

• Structured Illumination (SI) is crucial for recovering information like phase and shape.
• Three-Step Phase Shifting (TPS) is a common SI workflow but faces limitations in real-time applications.
• Existing systems suffer from low throughput, synchronization issues, and bulky implementations.

Purpose of the Study:

• To investigate, design, and validate a deterministic, low-latency, and portable projection-capture synchronization system for TPS-based SI.
• To overcome the limitations of current state-of-the-art (SOTA) SI systems.
• To enable faster and more efficient real-time SI acquisition.

Main Methods:

• Evaluated a Hyperspectral (HS) Python-based SI (HSPy-SI) system using a snapshot camera and fixed-delay synchronizer.
• Developed and prototyped the HyperSI system on heterogeneous embedded platforms (SBC and SoC).
• Introduced a tunable 'Frame Count to Wait' (W) parameter to optimize synchronization.

Main Results:

• HyperSI achieved over 8× higher pattern-generation throughput compared to SOTA.
• Polarized acquisition speed increased 7× (to nearly 4 FPS), and unpolarized acquisition reached ~12 FPS.
• The system demonstrated an 88× reduction in waiting time, significantly improving efficiency.

Conclusions:

• The proposed HyperSI system offers a deterministic, low-latency, and portable solution for TPS-based SI.
• HyperSI overcomes major limitations of real-time SI, enabling higher throughput and faster acquisition.
• This advancement facilitates more practical and widespread application of real-time SI techniques.