Parallelization of Dense Fluid Motion Estimation Application using OpenMP

Nitin Jain 1 Etienne Mémin 2 Christian Pérez 3
2 VISTA - Vision spatio-temporelle et active
IRISA - Institut de Recherche en Informatique et Systèmes Aléatoires, Inria Rennes – Bretagne Atlantique
3 PARIS - Programming distributed parallel systems for large scale numerical simulation
IRISA - Institut de Recherche en Informatique et Systèmes Aléatoires, ENS Cachan - École normale supérieure - Cachan, Inria Rennes – Bretagne Atlantique
Abstract : In this paper, we discuss the parallelization of a computer vision applicatio- n for dense fluid motion estimation using OpenMP. The application uses a recent algorithm based on energy-based motion estimator for predicting and analyzing the motion in image sequences showing fluid phenomenon. Standard techniques from computer vision are not well adapted for such images because of the great deal of spatial and temporal distortions in luminance patterns. The multiresolution multigrid framework of the application renders it amenable to parallelization after appropriate changes to the algorithm. We discuss why OpenMP is a suitable alternative to the conventional message passing model and compare it with other models based on various aspects. The results obtained on a SMP machine from different parallelization strategies are demonstrated and compared.
Document type :
Reports
Complete list of metadatas

https://hal.inria.fr/inria-00072032
Contributor : Rapport de Recherche Inria <>
Submitted on : Tuesday, May 23, 2006 - 7:36:33 PM
Last modification on : Tuesday, October 1, 2019 - 3:18:13 PM
Long-term archiving on: Sunday, April 4, 2010 - 8:57:31 PM

Identifiers

  • HAL Id : inria-00072032, version 1

Citation

Nitin Jain, Etienne Mémin, Christian Pérez. Parallelization of Dense Fluid Motion Estimation Application using OpenMP. [Research Report] RR-4556, INRIA. 2002. ⟨inria-00072032⟩

Share

Metrics

Record views

258

Files downloads

177