by Frank Davoine
Abstract:
The principle of fractal image coding presented in this paper is based on the theory of L-IFS (Local Iterated Function Systems). The algorithm exploits the fact that a real-world image is formed approximately of transformed copies of parts of itself. Thus, the construction of fractal codes is directly made on partitions of the image support. It is based on piece-wise similarities between blocks of different sizes.\\ The paper starts with a regular block based approach first proposed by Jacquin. To improve the algorithm, adaptive partitions are proposed with, in particular, the Delaunay triangulation. The results show an improvement in computing times, compression ratios and visual quality of reconstructed images.
Reference:
Fractal Image Compression Based on Adaptive Tessellations (Frank Davoine), Technical report, PRIP, TU Wien, 1993.
Bibtex Entry:
@TechReport{TR025,
author = "Frank Davoine",
institution = "PRIP, TU Wien",
number = "PRIP-TR-025",
title = "Fractal {I}mage {C}ompression {B}ased on {A}daptive
{T}essellations",
year = "1993",
url = "https://www.prip.tuwien.ac.at/pripfiles/trs/tr25.pdf",
abstract = "The principle of fractal image coding presented in
this paper is based on the theory of L-IFS (Local
Iterated Function Systems). The algorithm exploits
the fact that a real-world image is formed
approximately of transformed copies of parts of
itself. Thus, the construction of fractal codes is
directly made on partitions of the image support. It
is based on piece-wise similarities between blocks
of different sizes.\\ The paper starts with a
regular block based approach first proposed by
Jacquin. To improve the algorithm, adaptive
partitions are proposed with, in particular, the
Delaunay triangulation. The results show an
improvement in computing times, compression ratios
and visual quality of reconstructed images.",
}