Visualization of Astronomical Nebulae via Distributed Multi-GPU Compressed Sensing Tomography.

Area of Science:

• Astronomy and Astrophysics
• Computer Vision
• Scientific Visualization

Background:

• 3D visualization of astronomical nebulae is difficult due to limited 2D observational data.
• Existing methods struggle to capture the full detail and variety of nebulae.

Purpose of the Study:

• To develop a novel tomographic approach for generating realistic 3D nebulae visualizations.
• To improve the preservation of detail and visual variety in volumetric reconstructions.

Main Methods:

• Utilized a tomographic reconstruction technique exploiting nebulae symmetry (spherical/axial).
• Extended iterative compressed sensing with position-dependent regularization and linear equality constraints.
• Implemented a distributed multi-GPU system for high-resolution dataset reconstruction.

Main Results:

• Successfully generated plausible and realistic volumetric visualizations of nebulae.
• Demonstrated robustness and scalability using Hubble Space Telescope data.
• Achieved superior detail and visual variety compared to previous methods, especially for approximate symmetries.

Conclusions:

• The proposed tomographic method significantly enhances 3D nebulae visualization capabilities.
• The approach is robust, scalable, and effective for astronomical imagery.
• This technique offers a valuable tool for studying nebulae structure and evolution.