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

Online selection of discriminative tracking features.

Robert T Collins1, Yanxi Liu, Marius Leordeanu

  • 1Computer Science Engineering Department, The Pennsylvania State University, University Park, PA 16802, USA. rcollins@cse.psu.edu

IEEE Transactions on Pattern Analysis and Machine Intelligence
|October 22, 2005
PubMed
Summary

This study introduces an adaptive feature selection method for object tracking. It identifies features that best distinguish objects from backgrounds, improving tracking accuracy even with changing appearances.

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

Approval and delay status of United States -approved new drugs (2010-2023) in European Union, Japan, and China.

BMC health services research·2026
Same author

Correction to "Light-Induced Transformation from Covalent to Supramolecular Polymer Networks".

ACS macro letters·2026
Same author

Exploring factors influencing maternal willingness to donate breast milk: a descriptive qualitative study.

Frontiers in global women's health·2026
Same author

Estimated glucose disposal rate outperforms inflammation surrogate markers in predicting cardiometabolic multimorbidity among middle-aged and older adults.

Cardiovascular diabetology·2026
Same author

A Beta-Binomial Model for Estimating Zero- or One-inflated Pain Trajectories.

bioRxiv : the preprint server for biology·2026
Same author

Exploration of the signaling pathway involving OsCSN8 in regulating the leaf angle of rice.

TAG. Theoretical and applied genetics. Theoretische und angewandte Genetik·2026

Area of Science:

  • Computer Vision
  • Machine Learning
  • Pattern Recognition

Background:

  • Object tracking is crucial in computer vision.
  • Effective feature selection is key to robust tracking performance.
  • Adaptive methods are needed to handle variations in object appearance and background.

Purpose of the Study:

  • To develop an online feature selection mechanism for improving object tracking.
  • To identify and adaptively select features that best discriminate between objects and backgrounds.
  • To enhance tracking system robustness against appearance changes.

Main Methods:

  • Utilizing log likelihood ratios of class conditional sample densities to generate candidate features.
  • Employing the two-class variance ratio to rank features based on discriminative power.

Related Experiment Videos

  • Integrating the feature selection mechanism into a mean-shift tracking system for adaptive selection.
  • Developing a complementary approach to mitigate background clutter distraction.

    • Main Results:

    • The proposed method adaptively selects discriminative features for improved tracking.
    • Demonstrated ability to handle changing object appearances and background scenes.
    • Identified and addressed the susceptibility of variance ratio to background clutter.

    Conclusions:

    • Online feature selection based on object-background discrimination significantly enhances tracking performance.
    • The adaptive mechanism provides robustness to appearance variations.
    • The developed method offers a promising solution for real-world object tracking challenges.