Feasibility of a Real-Time Embedded Hyperspectral Compressive Sensing Imaging System

Area of Science:

• Optics and Photonics
• Computer Science
• Signal Processing

Background:

• Hyperspectral imaging (HSI) offers rich spectral data for diverse applications like remote sensing and medicine.
• Compressive HSI systems enable data acquisition with fewer samples than traditional methods.
• Reconstruction algorithms are crucial for recovering data in compressive HSI but are computationally intensive.

Purpose of the Study:

• To analyze the performance requirements for real-time compressive hyperspectral imaging.
• To identify strategies for accelerating HSI data reconstruction.
• To assess the feasibility of real-time applications for compressive HSI systems.

Main Methods:

• Analysis of computational power, memory, and bandwidth needs for a state-of-the-art reconstruction algorithm.
• Evaluation of algorithmic and hardware acceleration techniques (GPUs, FPGAs).
• Investigation of system matrix sparsity and data encoding for bandwidth reduction.

Main Results:

• Real-time reconstruction for compressive hyperspectral imaging is achievable.
• Exploiting system matrix sparsity significantly reduces computational load.
• Optimized data value encoding minimizes bandwidth requirements.

Conclusions:

• Compressive hyperspectral imaging can achieve real-time performance through combined acceleration strategies.
• Hardware acceleration and algorithmic optimizations are key to overcoming reconstruction bottlenecks.
• This work paves the way for wider adoption of compressive HSI in time-sensitive applications.