Advertisement

A new parallel MIMD Connected Component Labeling algorithm

  • Thomas Johansson
  • Ewert Bengtsson
Poster Session
Part of the Lecture Notes in Computer Science book series (LNCS, volume 817)

Abstract

A new parallel MIMD Connected Component Labeling algorithm of binary objects is presented. The binary image is sliced uniformly between the processors. Since objects can occur at the boundaries of each slice, interprocessor communication is needed to spread the information about which objects are touching the boundary. The essential difference to other previously described parallel labeling algorithms is in the way the proposed algorithm constructs the equivalence tables and how they are merged between the processor in the network. An iPSC/2 is used to verify the algorithm by empirical tests.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Embrechts H., Connected component labelling on a distributed memory multiprocessor, Hypercube and Distributed Computers, Elsevier S.P.B.V, (1989) 5–17Google Scholar
  2. 2.
    Perroton L., A hypercube algorithm for connected component labeling of volumic images, Research Report from the LIP,ENS, Lyon, Report nr 92-39, (1992)Google Scholar
  3. 3.
    Thurfjell L., Bengtsson E., Nordin B., A new three-dimensional Connected Component Labeling algorithm with simultaneous object feature extraction capability, Graphical Models and Image Processing, Vol 54, No 4 (1992) 357–364CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1994

Authors and Affiliations

  • Thomas Johansson
    • 1
  • Ewert Bengtsson
    • 1
  1. 1.Centre for Image AnalysisUppsala UniversitySweden

Personalised recommendations