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

Low-level aspects of segmentation and recognition.

S Ullman1

  • 1Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge 02139.

Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences
|September 29, 1992
PubMed
Summary

This study presents novel methods for 3D object recognition, focusing on image segmentation using contour geometry and object recognition via 2D view combinations. These low-level approaches enable robust 3D object identification without explicit 3D models.

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

Circulating microparticles and plasma levels of soluble E- and P-selectins in patients with systemic sclerosis.

Scandinavian journal of rheumatology·2013
Same author

Clinical prediction of 5-year survival in systemic sclerosis: validation of a simple prognostic model in EUSTAR centres.

Annals of the rheumatic diseases·2011
Same author

Skin signs in Ehlers-Danlos syndrome: clinical tests and para-clinical methods.

Scandinavian journal of rheumatology·2010
Same author

A randomized controlled trial of R-salbutamol for topical treatment of discoid lupus erythematosus.

The British journal of dermatology·2009
Same author

Skin extensibility and consistency in patients with Ehlers-Danlos syndrome and benign joint hypermobility syndrome.

Scandinavian journal of rheumatology·2009
Same author

Factors affecting survival in juvenile systemic sclerosis.

Rheumatology (Oxford, England)·2008

Area of Science:

  • Computer Vision
  • Artificial Intelligence
  • Robotics

Background:

  • Accurate three-dimensional (3D) object recognition is crucial for applications like robotics and autonomous systems.
  • Traditional methods often rely on complex 3D models, which can be computationally intensive and difficult to acquire.
  • Challenges include segmenting relevant objects from complex scenes and recognizing them from various viewpoints.

Purpose of the Study:

  • To develop low-level, computationally efficient methods for 3D object recognition.
  • To address the challenges of image segmentation and viewpoint variation in 3D object recognition.
  • To enable object recognition without the need for explicit 3D models.

Main Methods:

  • Segmentation: Extraction of globally salient structures from contour images using geometrical attributes (smoothness, contour length) via a parallel network of neuron-like elements.

Related Experiment Videos

  • Recognition: Overcoming viewpoint dependency by using linear combinations of a limited set of 2D object views.
  • Overall Approach: Emphasis on bottom-up processing driven by image contour geometry and direct manipulation of 2D images.

    • Main Results:

    • Demonstrated effective segmentation of salient structures based on intrinsic image properties.
    • Showcased a method for viewpoint-invariant recognition using 2D image representations.
    • Achieved 3D object recognition through low-level image processing techniques.

    Conclusions:

    • The proposed methods offer efficient and robust solutions for 3D object recognition tasks.
    • The bottom-up segmentation and 2D view-based recognition approach bypasses the need for explicit 3D models.
    • These techniques contribute to advancing computer vision and AI in handling real-world object recognition challenges.