Computer Graphics
TU Braunschweig

3D Reconstruction of Reflection Nebulae from a Single Image

3D Reconstruction of Reflection Nebulae from a Single Image

This paper presents a method for reconstructing the 3D distribution of dust densities in reflection nebulae based on a single input image using an analysisby-synthesis approach. In a reflection nebula, light is typically emitted from a central star and then scattered and partially absorbed by the nebula’s dust particles. We model the light transport in this kind of nebulae by considering absorption and single

scattering only. While the core problem of reconstructing an arbitrary 3D volume of dust particles

from a 2D image would be ill-posed we demonstrate how the special configuration of light transport

paths in reflection nebulae allows us to produce non-exact but plausible 3D volumes. Our reconstruction is driven by an iterative non-linear optimization method, which renders an image in each step with the current estimate of dust densities and then updates the density values to minimize the error to the input image. The recovered volumetric datasets can be used in astrophysical research as well as planetarium visualizations.

Author(s):Andrei Lintu, Lars Hoffmann, Marcus Magnor, Hendrik Lensch, Hans-Peter Seidel
Published:November 2007
Type:Article in conference proceedings
Book:Proc. Vision, Modeling and Visualization (VMV)
Presented at:Vision, Modeling and Visualization (VMV) 2007
Project(s): Astrophysical Modeling and Visualization 

  title = {3D Reconstruction of Reflection Nebulae from a Single Image},
  author = {Lintu, Andrei and Hoffmann, Lars and Magnor, Marcus and Lensch, Hendrik and Seidel, Hans-Peter},
  booktitle = {Proc. Vision, Modeling and Visualization ({VMV})},
  organization = {Eurographics},
  pages = {109--116},
  month = {Nov},
  year = {2007}