DocumentCode
937374
Title
Dominance--A relation between distortion measures
Author
Stjernvall, Jan-Erik
Volume
29
Issue
6
fYear
1983
fDate
11/1/1983 12:00:00 AM
Firstpage
798
Lastpage
807
Abstract
A relation between distortion measures for source coding, called dominance, is defined so that if
and
are two distortion measures and
dominates
, then a source coding system which meets the distortion criterion
simply yields a system meeting the distortion criterion
by a simple change of decoder. It is shown that a very simple characterization of dominance can be given in terms of the characteristic sets of the distortion measures. The characteristic set of a distortion matrix is defined as the sum of the convex hull of the columns of the matrix and the nonnegative orthant. If two distortion measures have the same characteristic set they are regarded as equivalent. It is shown that the definition of dominance is robust in the sense that some alternative definitions are equivalent to the given one. The definition of dominance is extended to the case where a pair of distortion measures dominates a third distortion measure. Here a simple characterization is given in terms of the characteristic sets of the distortion measures.
and
are two distortion measures and
dominates
, then a source coding system which meets the distortion criterion
simply yields a system meeting the distortion criterion
by a simple change of decoder. It is shown that a very simple characterization of dominance can be given in terms of the characteristic sets of the distortion measures. The characteristic set of a distortion matrix is defined as the sum of the convex hull of the columns of the matrix and the nonnegative orthant. If two distortion measures have the same characteristic set they are regarded as equivalent. It is shown that the definition of dominance is robust in the sense that some alternative definitions are equivalent to the given one. The definition of dominance is extended to the case where a pair of distortion measures dominates a third distortion measure. Here a simple characterization is given in terms of the characteristic sets of the distortion measures.Keywords
Rate-distortion theory; Source coding; Decoding; Distortion measurement; Information theory; Probability distribution; Random variables; Robustness; Source coding;
fLanguage
English
Journal_Title
Information Theory, IEEE Transactions on
Publisher
ieee
ISSN
0018-9448
Type
jour
DOI
10.1109/TIT.1983.1056758
Filename
1056758
Link To Document