by Horst Bischof, Walter G. Kropatsch
Abstract:
Neural networks and image pyramids are massively parallel processing structures. In this paper we exploit the similarities as well as the differences between these structures. The general goal is to exchange knowledge between these two fields. After introducing the basic concepts of neural networks and image pyramids we give a translation table of the vocabulary used in image pyramids and those used in neural networks. In the following sections we compare neural networks and image pyramids in detail. We show how a modified Hopfield network can be used for irregular decimation. We examine the type of knowledge stored and the processing performed by pyramids and neural networks. In the case of numerical information, so called 'numerical pyramids' are rather similar to neural networks. But also for 'symbolic pyramids' we show how to implement them by neural networks. In particular we present a neural implementation of the 2x2/2 curve pyramid. We derive some general rules for implementing symbolic pyramids by neural networks. Finally we briefly discuss the role of learning in image pyramids.
Reference:
Neural Networks versus Image Pyramids (Horst Bischof, Walter G. Kropatsch), Technical report, PRIP, TU Wien, 1993.
Bibtex Entry:
@TechReport{TR007,
author = "Horst Bischof and Walter G. Kropatsch",
institution = "PRIP, TU Wien",
number = "PRIP-TR-007",
title = "Neural {N}etworks versus {I}mage {P}yramids",
year = "1993",
url = "https://www.prip.tuwien.ac.at/pripfiles/trs/tr7.pdf",
abstract = "Neural networks and image pyramids are massively
parallel processing structures. In this paper we
exploit the similarities as well as the differences
between these structures. The general goal is to
exchange knowledge between these two fields. After
introducing the basic concepts of neural networks
and image pyramids we give a translation table of
the vocabulary used in image pyramids and those used
in neural networks. In the following sections we
compare neural networks and image pyramids in
detail. We show how a modified Hopfield network can
be used for irregular decimation. We examine the
type of knowledge stored and the processing
performed by pyramids and neural networks. In the
case of numerical information, so called 'numerical
pyramids' are rather similar to neural networks. But
also for 'symbolic pyramids' we show how to
implement them by neural networks. In particular we
present a neural implementation of the 2x2/2 curve
pyramid. We derive some general rules for
implementing symbolic pyramids by neural
networks. Finally we briefly discuss the role of
learning in image pyramids.",
}