Invariant pattern recognition using multiple filter image representations
by ,
Abstract:
In this paper we develop a 4-dimensional representation for patterns based on image decompositions via orientation- and size-specific filters. By retaining image positional information, this encoding scheme reduces pattern rotations, translations, and scale changes to shifts in the filter outputs. The appropriate correlation processes for matching are discussed and the recognition system is illustrated by a number of examples.
Reference:
Invariant pattern recognition using multiple filter image representations (Christoph Zetzsche, Terry Caelli), In Computer Vision, Graphics, and Image Processing, Elsevier, volume 45, 1989.
Bibtex Entry:
@Article{Zetzsche1989,
  author    = {Christoph Zetzsche and Terry Caelli},
  title     = {Invariant pattern recognition using multiple filter image representations},
  journal   = {Computer Vision, Graphics, and Image Processing},
  year      = {1989},
  volume    = {45},
  number    = {2},
  pages     = {251--262},
  month     = {feb},
  abstract  = {In this paper we develop a 4-dimensional representation for patterns based on image decompositions via orientation- and size-specific filters. By retaining image positional information, this encoding scheme reduces pattern rotations, translations, and scale changes to shifts in the filter outputs. The appropriate correlation processes for matching are discussed and the recognition system is illustrated by a number of examples.},
  doi       = {10.1016/0734-189x(89)90135-7},
  publisher = {Elsevier},
  url       = {http://dx.doi.org/10.1016/0734-189X(89)90135-7},
}