DI-UMONS : Dépôt institutionnel de l’université de Mons

Recherche transversale
Rechercher
(titres de publication, de périodique et noms de colloque inclus)
2014-04-07 - Livre/Chapitre ou partie - Anglais - 22 page(s)

Mahmoudi Sidi , Ozkan Erencan, Manneback Pierre , Tosun Souleyman, "Taking Advantage of Heterogeneous Platforms in Image and Video Processing" in "Complex HPC book" , 978-1-118-71205-4

  • Edition : Wiley
  • Codes CREF : Techniques d'imagerie et traitement d'images (DI2770), Technologie informatique hardware (DI2560), Informatique appliquée logiciel (DI2570), Informatique générale (DI1162)
  • Unités de recherche UMONS : Informatique, Logiciel et Intelligence artificielle (F114)
  • Instituts UMONS : Institut de Recherche en Technologies de l’Information et Sciences de l’Informatique (InforTech), Institut NUMEDIART pour les Technologies des Arts Numériques (Numédiart)
  • Centres UMONS : Centre de Recherche en Technologie de l’Information (CRTI)

Abstract(s) :

(Anglais) Image and video processing algorithms present a necessary tool for various domains related to computer vision such as pattern recognition, video surveillance, medical imaging, etc. These algorithms become so hampered by their high consumption of computing times when processing large sets of high resolution images and videos. In this work, we propose a model enabling an efficient exploitation of parallel (GPU) and heterogeneous (Multi-CPU/Multi-GPU) platforms, in order to improve performance. The proposed model enables an efficient scheduling of hybrid tasks and an effective management of heterogeneous memories. Based on this model, we developed hybrid implementations of edge and corner detection methods which were exploited in a medical application of vertebra detection. We developed also GPU implementations of several video processing algorithms such as background subtraction, silhouette extraction, optical flow computation which were exploited for accelerating an application of real time camera motion estimation. Experimental results showed a global speedup ranging from 5 to 40, when processing different sets of images, by comparison with CPU implementations.