by Dieter Willersinn
Abstract:
Hierarchical representation of images is a crucial building principle of a system architecture that can cope with the complexity of visual perception. This paper presents an algorithm that builds a hierarchy which is flexible enough to provide a shift and scale invariant abstract representation of image content. The hierarchy is irregular in the sense that an element of the representation may have an arbitrary number of neighbors. At the same time, the representation can be described by a bounded data structure. Main properties of the algorithm with respect to visual perception are proved.
Reference:
Parallel Graph Contraction for Dual Irregular Pyramids (Dieter Willersinn), Technical report, PRIP, TU Wien, 1994.
Bibtex Entry:
@TechReport{TR028,
author = "Dieter Willersinn",
institution = "PRIP, TU Wien",
number = "PRIP-TR-028",
title = "Parallel {G}raph {C}ontraction for {D}ual
{I}rregular {P}yramids",
year = "1994",
url = "https://www.prip.tuwien.ac.at/pripfiles/trs/tr28.pdf",
abstract = "Hierarchical representation of images is a crucial
building principle of a system architecture that can
cope with the complexity of visual perception. This
paper presents an algorithm that builds a hierarchy
which is flexible enough to provide a shift and
scale invariant abstract representation of image
content. The hierarchy is irregular in the sense
that an element of the representation may have an
arbitrary number of neighbors. At the same time, the
representation can be described by a bounded data
structure. Main properties of the algorithm with
respect to visual perception are proved.",
}