Adding parallelism to the hybrid image processing library in multi-threading and multi-core systems

  • Authors:
  • Boguslaw Cyganek

  • Affiliations:
  • AGH University of Science and Technology, Al. Mickiewicza 30, Krakow 30-059, Poland

  • Venue:
  • NESEA '11 Proceedings of the 2011 IEEE 2nd International Conference on Networked Embedded Systems for Enterprise Applications
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

The paper presents a hybrid image processing library adapted to take full advantage of the multi-threading and multi-core computer systems. Architecture of the library follows the handle-body design pattern. The handle layer represents an abstract interface which hides implementation details and leaves the user's code untouched. Software/hardware implementations are hidden in the body layers which allow for high level of parallelism and best exploitation of the available resources. The first aim of this paper is to present a short introduction to the new version of the library which is available from the Internet. The second purpose is to present refactoring of the software layer with help of the OpenMP to take advantage of the multi-core systems, as well as to share expertise in using this technology in other object-oriented platforms for massive data processing. Experimental results show on average two-three times acceleration in the refactored layers.