A fast parallel method for synthesis of random patterns

Peter Meer, Steven Connelly

Research output: Contribution to journalArticlepeer-review

12 Scopus citations

Abstract

A new method for generation of nonhomogeneous, anisotropic random patterns defined on a square lattice is presented. First a hierarchy of seed structures is derived by recursively deleting the sites of the square lattice. This decimation process is parallel, and employs local ordering of i.i.d. random numbers. At every level of the hierarchy the surviving seeds expand and incorporate the decimated sites into cells. The ensemble of cells down-projected onto the square lattice defines a random tessellation whose coarseness increases with the level in the hierarchy. Nonhomogeneous random patterns are obtained by concatenating cells taken from different levels; anisotropic patterns by employing unequal weights in the generation process. The underlying hierarchical structure assures that the random patterns are synthesised in O[log(image_size)]time.

Original languageEnglish (US)
Pages (from-to)189-204
Number of pages16
JournalPattern Recognition
Volume22
Issue number2
DOIs
StatePublished - 1989

All Science Journal Classification (ASJC) codes

  • Software
  • Signal Processing
  • Computer Vision and Pattern Recognition
  • Artificial Intelligence

Keywords

  • Image pyramids
  • Parallel processing
  • Random patterns

Fingerprint Dive into the research topics of 'A fast parallel method for synthesis of random patterns'. Together they form a unique fingerprint.

Cite this