VCLab

 

RESEARCH AREAS   PEOPLE   PUBLICATIONS   COURSES   ABOUT US
Home / Publications

indicator

Computer Vision and Pattern Recognition (CVPR 2024)

 
OmniSDF: Scene Reconstruction
using Omnidirectional Signed Distance Functions and Adaptive Binoctrees
 
  Hakyeong Kim Andreas Meuleman Hyeonjoong Jang James Tompkin Min H. Kim  
  KAIST INRIA KAIST Brown University KAIST  
 
 
  We introduce a memory-efficient neural 3D reconstruction method tailored to work with short egocentric omnidirectional video inputs. The geometry is estimated using a signed distance field and a novel adaptive spherical binoctree data structure subdivided through iterative optimization. We show that our method outperforms other state-of-the-art 3D reconstruction methods in balancing detail and memory cost.  
     
   
  Supplemental video
     
   
  CVPR 2024 Presentation
   
  Abstract
   
 

We present a method to reconstruct indoor and outdoor static scene geometry and appearance from an omnidirectional video moving in a small circular sweep. This setting is challenging because of the small baseline and large depth ranges, making it difficult to find ray crossings. To better constrain the optimization, we estimate geometry as a signed distance field within a spherical binoctree data structure and use a complementary efficient tree traversal strategy based on a breadth-first search for sampling. Unlike regular grids or trees, the shape of this structure well-matches the camera setting, creating a better memory-quality trade-off. From an initial depth estimate, the binoctree is adaptively subdivided throughout the optimization; previous methods use a fixed depth that leaves the scene undersampled. In comparison with three neural optimization methods and two non-neural methods, ours shows decreased geometry error on average, especially in a detailed scene, while significantly reducing the required number of voxels to represent such details.

     
   
  BibTeX
 
@InProceedings{Kim_2024_CVPR,
   author = {Hakyeong Kim and Andreas Meuleman and Hyeonjoong Jang and 
   James Tompkin and Min H. Kim},
   title = {OmniSDF: Scene Reconstruction using
   Omnidirectional Signed Distance Functions and Adaptive Binoctrees},
   booktitle = {IEEE Conference on Computer Vision and 
      Pattern Recognition (CVPR)},
   month = {June},
   year = {2024}
} 
   
   
icon
Preprint paper:
PDF (6.6MB)
icon
Supplemental
material #1:
PDF (5.3MB)
www GitHub
code
 

Hosted by Visual Computing Laboratory, School of Computing, KAIST.

KAIST