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

Video content classification based on 3-D eigen analysis.

Jie Wei1

  • 1Department of Computer Science, City College of New York, New York, NY 10031, USA. wei@cs.ccny.cuny.edu

IEEE Transactions on Image Processing : a Publication of the IEEE Signal Processing Society
|May 13, 2005
PubMed
Summary

This study introduces a novel 3D eigen analysis for efficient video classification. The method treats videos as 3D solids, effectively utilizing spatial and temporal features for improved content understanding.

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

[Close reduction for the treatment of inferior intra-articular dislocation of patella:a case report and literature review].

Zhongguo gu shang = China journal of orthopaedics and traumatology·2019
Same author

Analgesic efficacy and safety of ketamine after total knee or hip arthroplasty: a meta-analysis of randomised placebo-controlled studies.

BMJ open·2019
Same author

In situ Spectroscopic Insight into the Origin of the Enhanced Performance of Bimetallic Nanocatalysts towards the Oxygen Reduction Reaction (ORR).

Angewandte Chemie (International ed. in English)·2019
Same author

A novel method for estimating nail-tract bone density for intertrochanteric fractures.

Journal of orthopaedic translation·2019
Same author

Endophilin A2 attenuates cardiac hypertrophy induced by isoproterenol through the activation of autophagy.

American journal of translational research·2019
Same author

Expression and clinical significance of CPS1 in glioblastoma multiforme.

Current research in translational medicine·2019

Area of Science:

  • Computer Vision
  • Machine Learning
  • Signal Processing

Background:

  • Video understanding and classification are crucial for many applications.
  • Existing methods often treat spatial and temporal features separately.
  • Eigen analysis has shown success in still-image analysis.

Purpose of the Study:

  • To develop an efficient and effective content-based video classification method.
  • To leverage both spatial and temporal features simultaneously.
  • To adapt eigen analysis for video data.

Main Methods:

  • Developed a video classification method based on three-dimensional (3-D) eigen analysis.
  • Treated videos as solids in a 3-D Euclidean space.
  • Computed eigen values and vectors of autocorrelation matrices for 3-D macroblocks.

Related Experiment Videos

  • Assigned labels based on the properties of eigen values and vectors.

    • Main Results:

    • The proposed 3-D eigen analysis method demonstrated encouraging performance.
    • The approach effectively integrates spatial and temporal information.
    • Empirical studies validated the efficacy of the method.

    Conclusions:

    • 3-D eigen analysis offers a promising approach for video classification.
    • Treating videos as 3D entities enhances feature utilization.
    • This method provides an efficient and effective solution for video content understanding.