Enforcing stochastic inverse consistency in non-rigid image registration and matching

S.K. Yeung, C.K. Tang, P. Shi, J.P.W. Pluim, M.A. Viergever, A.C S Chung, H.C. Shen

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

10 Citations (Scopus)

Abstract

This paper presents a new method to enforce inverse consistency in nonrigid image registration and matching. Conventional approaches assume diffeomorphic transformation, implicitly or explicitly. However, the inherent smoothness constraint discourages discontinuity consideration. We propose a post-processing algorithm that integrates the input forward and backward fields, which are output by existing registration/matching algorithms, to produce more robust results. Given such a pair of input fields, our algorithm alternately refines the fields by tensor belief propagation, and enforces inverse consistency in stochastic sense by generalized total least squares fitting. To show the efficacy of our stochastic inverse consistency approach, we first present results on very noisy fields. We then demonstrate improvement on existing stereo matching where occlusion is naturally handled by localizing violations of inverse consistency. Finally, we propose a novel application on image stitching, where stochastic inverse consistency is employed in structure deformation, in order to seamlessly align overlapping images with severe misalignment in structure and intensity.

Original languageEnglish
Title of host publication26th IEEE Conference on Computer Vision and Pattern Recognition, CVPR, 23-28 June 2008, Anchorage, Arkansas
Place of PublicationPiscataway
PublisherInstitute of Electrical and Electronics Engineers
Number of pages8
ISBN (Print)978-1-4244-2242-5
DOIs
Publication statusPublished - 2008
Externally publishedYes
Event26th IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2008) - Anchorage, United States
Duration: 23 Jun 200828 Jun 2008
Conference number: 26

Conference

Conference26th IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2008)
Abbreviated titleCVPR 2008
CountryUnited States
CityAnchorage
Period23/06/0828/06/08

Keywords

  • Image registration and matching

Fingerprint Dive into the research topics of 'Enforcing stochastic inverse consistency in non-rigid image registration and matching'. Together they form a unique fingerprint.

Cite this