Compression and denoising of time-resolved light transport

Area of Science:

• Optics and Photonics
• Computer Vision
• Data Science

Background:

• Ultra-fast imaging captures light propagation for advanced applications like hidden geometry reconstruction and seeing through scattering media.
• High-dimensional, high-resolution light transport data poses significant performance and storage challenges.
• Noise in captured and synthesized data degrades signal quality over time, impacting results.

Purpose of the Study:

• To develop a method for efficiently representing time-resolved light transport data.
• To address performance and storage constraints associated with high-dimensional temporal data.
• To mitigate signal degradation caused by noise in light transport data.

Main Methods:

• Feature extraction to create meaningful representations of time-resolved light transport.
• Data compression techniques applied to temporal light transport data.
• Variance reduction in both temporal and spatial dimensions of the data.

Main Results:

• Reduced size of time-resolved transport data by up to a factor of 32.
• Significant mitigation of variance in temporal and spatial dimensions.
• Accurate representation of time-resolved light transport data achieved.

Conclusions:

• The proposed method effectively addresses challenges in handling time-resolved light transport data.
• Significant data size reduction and noise mitigation are achieved.
• Enables more efficient and robust applications relying on ultra-fast light propagation imaging.