Compressed sensing for active non-line-of-sight imaging

Area of Science:

• Optics and Photonics
• Computational Imaging
• Signal Processing

Background:

• Non-line-of-sight (NLOS) imaging allows "looking around corners," a capability with expanding applications.
• Active NLOS imaging utilizes light or other signals to reconstruct hidden scenes.
• Traditional NLOS methods can require extensive data acquisition, limiting speed.

Purpose of the Study:

• To investigate the efficacy of compressed sensing in active NLOS imaging.
• To determine if compressed sensing can reduce data acquisition requirements without compromising image quality.
• To analyze the performance of compressed sensing in both confocal NLOS imaging and active occlusion-based periscopy.

Main Methods:

• Applying compressed sensing principles to active NLOS imaging scenarios.
• Analyzing theoretical reductions in scanning points required for reconstruction.
• Conducting experimental validation using confocal NLOS imaging.

Main Results:

• Compressed sensing substantially reduces the number of scanning points needed for NLOS imaging.
• High-resolution 3D hidden images (64x64 spatial resolution) were reconstructed using minimal scanning points (5x5).
• Imaging quality was maintained despite the significant reduction in data acquisition.

Conclusions:

• Compressed sensing is a viable technique for enhancing active NLOS imaging efficiency.
• Reduced scanning points and capture time, coupled with preserved imaging quality, make this approach suitable for high-speed applications.
• This method offers a significant advancement for practical NLOS imaging systems.