An efficient people tracking system using the FFT-Correlation and Optimized SAD

TitreAn efficient people tracking system using the FFT-Correlation and Optimized SAD
Publication TypeJournal Article
Year of Publication2013
AuthorsEzzahout, Aa, Hadi, MYb, Thami, ROHa
JournalJournal of Theoretical and Applied Information Technology
Volume48
Pagination9-14
Abstract

The research questions which this study addresses are resumed in obtaining the picks of Optimized SAD and FFT-Correlation function. These values allow us to detect and track the blocs corresponding to the moving person. This Bloc-Matching system of our tracked bloc 8x8, simplify the process of detecting and tracking all the representative matrix of foreground. Practically, in this correspondence, we propose an implementation technique for detection and tracking system which detect and track people's in motion using a fixed single camera. These bi-techniques simplify the process of people tracking and give a new way for people tracking. Moreover, in this system we choose to detect moving people with the more robust algorithm (Gaussian Mixture Model) in term of misclassified pixels. Thereafter, we track a bloc 8x8 from the bounding box containing the moving person by returning the position of the similar bloc in the next frame. This bloc position is detected by maximization of the Fast Fourier Transform-Correlation (FFT-Corr) and minimizing the error: optimized Sum of Absolute Difference (OSAD) between the bloc reference and the similar bloc. © 2005 - 2013 JATIT & LLS. All rights reserved.

URLhttps://www.scopus.com/inward/record.uri?eid=2-s2.0-84876021321&partnerID=40&md5=cfb2300d8d0e312b0c77d271b683b278
Revues: 

Partenaires

Localisation


Location map

Suivez-nous sur

  

Contactez-nous

ENSIAS

Avenue Mohammed Ben Abdallah Regragui, Madinat Al Irfane, BP 713, Agdal Rabat, Maroc

Résultat de recherche d'images pour "icone fax" Télécopie : (+212) 5 37 77 72 30

    Compteur de visiteurs:280,151
    Education - This is a contributing Drupal Theme
    Design by WeebPal.