Home / Publications


British Machine Vision Conference (BMVC)

Urban Image Stitching using Planar Perspective Guidance
  Joo Ho Lee Seung-Hwan Baek Min H. Kim  
  Korea Advanced Institute of Science and Technology (KAIST)  
  Our image stitching method accounts for planar perspective while calculating local projective warp with the help of grid-like structural characteristics of urban scenes. It allows us to overcome wavy artifacts commonly observed by local homography-based warp approaches such as the As-Projective-As-Possible (APAP) method.  
  Image stitching methods with spatially-varying homographies have been proposed to overcome partial misalignments caused by global perspective projection; however, local warp operators often fracture the coherence of linear structures, resulting in an inconsistent perspective. In this paper, we propose an image stitching method that warps a source image to a target image by local projective warps using planar perspective guidance. We first detect line structures that converge into three vanishing points, yielding line-cluster probability functions for each vanishing point. Then we estimate local homographies that account for planar perspective guidance from the joint probability of planar guidance, in addition to spatial coherence. This allows us to enhance linear perspective structures while warping multiple urban images with grid-like structures. Our results validate the effectiveness of our method over state-of-the-art projective warp methods in terms of planar perspective.
  author  = {Joo Hoo Lee and Seung-Hwan Baek and Min H. Kim},
  title   = {Multiview Image Completion with Space Structure Propagation},
  booktitle = {Proc. British Machine Vision Conference (BMVC 2017)},
  address = {London, England},
  year = {2017},
  pages = {1-11},

Preprint paper:
Full-res. PDF

Supplemental materials:
a PDF file

BMVC 2017

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