Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Affine structure from motion.

J J Koenderink1, A J van Doorn

  • 1Buys Ballot Laboratory, Utrecht University, The Netherlands.

Journal of the Optical Society of America. A, Optics and Image Science
|February 1, 1991
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Method of stabilizing the retinal image.

Applied optics·2010
Same author

[An animal in need: trespassers are strictly handled].

Tijdschrift voor diergeneeskunde·2009
Same author

Spatiotemporal contrast detection threshold surface is bimodal.

Optics letters·2009
Same author

Do reference surfaces influence exocentric pointing?

Acta psychologica·2008
Same author

Optical properties (bidirectional reflectance distribution function) of shot fabric.

Applied optics·2008
Same author

Surface roughness from highlight structure.

Applied optics·2008
Same journal

Phase retrieval with prior information.

Journal of the Optical Society of America. A, Optics and image science·2008
Same journal

Clinical microscopy of the cornea utilizing optical sectioning and a high-numerical-aperture objective.

Journal of the Optical Society of America. A, Optics and image science·1993
Same journal

Eye-tracking laser Doppler velocimeter stabilized in two dimensions: principle, design, and construction.

Journal of the Optical Society of America. A, Optics and image science·1993
Same journal

Effects of aging in retinal image quality.

Journal of the Optical Society of America. A, Optics and image science·1993
Same journal

Axial eye-length measurement by wavelength-shift interferometry.

Journal of the Optical Society of America. A, Optics and image science·1993
Same journal

Fractal analysis of steady-state-flicker visual evoked potentials: feasibility.

Journal of the Optical Society of America. A, Optics and image science·1993
See all related articles

Inferring 3D structure from 2D images is challenging. This study reveals how humans process visual information, using fewer views and relaxing rigidity constraints for structure-from-motion tasks.

Area of Science:

  • Computer Vision
  • Computational Neuroscience
  • Geometric Deep Learning

Background:

  • The structure-from-motion problem aims to reconstruct 3D spatial configurations from 2D image sequences.
  • While three orthographic views suffice for rigid transformations, human vision utilizes fewer views effectively.

Purpose of the Study:

  • To investigate how spatial structure can be inferred from limited visual data, mimicking human capabilities.
  • To stratify solutions to the structure-from-motion problem based on assumptions and required geometric expertise.

Main Methods:

  • Developed a stratified approach to structure-from-motion, identifying distinct stages of geometric inference.
  • Introduced an affine stage that assumes smooth deformation, bypassing rigidity constraints and metric requirements.

Related Experiment Videos

  • Incorporated metrical methods for shape recovery and utilized a third view for calibration.

    • Main Results:

    • An affine stage allows inferring spatial configuration (modulo affinity) from just two views.
    • Metrical methods enable shape recovery from two views, up to depth scaling and shear.
    • A third view is sufficient to resolve calibration ambiguities.

    Conclusions:

    • Human visual perception of 3D structure from motion is more flexible than traditional rigid models suggest.
    • A stratified approach offers a clearer understanding of the assumptions and expertise needed for structure-from-motion.
    • The proposed method provides a framework for inferring 3D shape from limited, non-rigid visual input.