Adaptive Modeling and Coding for the Lossless Compression of Binary Images

Hidetoshi YOKOO; Ken SUDOH; Kazunori UEHARA

Adaptive Modeling and Coding for the Lossless Compression of Binary Images

Hidetoshi YOKOO, Ken SUDOH, Kazunori UEHARA

Full Text Views

0

Cite this

Summary :

This paper discusses lossless encoding methods of compressing two-dimensional black and white images for the efficient communication of such images. All the models concerned are adaptive in the sense that we assume no prior knowledge of statistical properties of any input and that the image is serially scanned and encoded in one pass. In the adaptive strategy, each picture element (pel) is encoded based on the time-variant coding parameters, which are empirically estimated from the already encoded pels in the present image. Since the coding parameters depend on the conditioning neighboring pels, modeling involves a pel-by-pel determination of conditioning neighboring pels. Two approximate solutions of how to select the conditioning pels are given, both of which are based on Rissanen's MDL criterion. Although no rigorous theory to account for the performance has been established, the experimental study demonstrates that one of them is superior in compression performance to any adaptive model utilizing a fixed number of conditioning pels. In this paper, an efficient implementation of models by means of an arithmetic code is also presented.

Publication: IEICE TRANSACTIONS on transactions Vol.E73-E No.10 pp.1640-1646

Publication Date: 1990/10/25

Publicized

Online ISSN

DOI

Type of Manuscript: Special Section PAPER (Special Issue on Communication Theory and Its Applications)

Category: Signal Processing and Image Compression

Authors

Hidetoshi YOKOO
Ken SUDOH
Kazunori UEHARA

Keyword

Cite this

Copy

Hidetoshi YOKOO, Ken SUDOH, Kazunori UEHARA, "Adaptive Modeling and Coding for the Lossless Compression of Binary Images" in IEICE TRANSACTIONS on transactions, vol. E73-E, no. 10, pp. 1640-1646, October 1990, doi: .
Abstract: This paper discusses lossless encoding methods of compressing two-dimensional black and white images for the efficient communication of such images. All the models concerned are adaptive in the sense that we assume no prior knowledge of statistical properties of any input and that the image is serially scanned and encoded in one pass. In the adaptive strategy, each picture element (pel) is encoded based on the time-variant coding parameters, which are empirically estimated from the already encoded pels in the present image. Since the coding parameters depend on the conditioning neighboring pels, modeling involves a pel-by-pel determination of conditioning neighboring pels. Two approximate solutions of how to select the conditioning pels are given, both of which are based on Rissanen's MDL criterion. Although no rigorous theory to account for the performance has been established, the experimental study demonstrates that one of them is superior in compression performance to any adaptive model utilizing a fixed number of conditioning pels. In this paper, an efficient implementation of models by means of an arithmetic code is also presented.
URL: https://global.ieice.org/en_transactions/transactions/10.1587/e73-e_10_1640/_p

Copy

@ARTICLE{e73-e_10_1640,
author={Hidetoshi YOKOO, Ken SUDOH, Kazunori UEHARA, },
journal={IEICE TRANSACTIONS on transactions},
title={Adaptive Modeling and Coding for the Lossless Compression of Binary Images},
year={1990},
volume={E73-E},
number={10},
pages={1640-1646},
abstract={This paper discusses lossless encoding methods of compressing two-dimensional black and white images for the efficient communication of such images. All the models concerned are adaptive in the sense that we assume no prior knowledge of statistical properties of any input and that the image is serially scanned and encoded in one pass. In the adaptive strategy, each picture element (pel) is encoded based on the time-variant coding parameters, which are empirically estimated from the already encoded pels in the present image. Since the coding parameters depend on the conditioning neighboring pels, modeling involves a pel-by-pel determination of conditioning neighboring pels. Two approximate solutions of how to select the conditioning pels are given, both of which are based on Rissanen's MDL criterion. Although no rigorous theory to account for the performance has been established, the experimental study demonstrates that one of them is superior in compression performance to any adaptive model utilizing a fixed number of conditioning pels. In this paper, an efficient implementation of models by means of an arithmetic code is also presented.},
keywords={},
doi={},
ISSN={},
month={October},}

Copy

TY - JOUR
TI - Adaptive Modeling and Coding for the Lossless Compression of Binary Images
T2 - IEICE TRANSACTIONS on transactions
SP - 1640
EP - 1646
AU - Hidetoshi YOKOO
AU - Ken SUDOH
AU - Kazunori UEHARA
PY - 1990
DO -
JO - IEICE TRANSACTIONS on transactions
SN -
VL - E73-E
IS - 10
JA - IEICE TRANSACTIONS on transactions
Y1 - October 1990
AB - This paper discusses lossless encoding methods of compressing two-dimensional black and white images for the efficient communication of such images. All the models concerned are adaptive in the sense that we assume no prior knowledge of statistical properties of any input and that the image is serially scanned and encoded in one pass. In the adaptive strategy, each picture element (pel) is encoded based on the time-variant coding parameters, which are empirically estimated from the already encoded pels in the present image. Since the coding parameters depend on the conditioning neighboring pels, modeling involves a pel-by-pel determination of conditioning neighboring pels. Two approximate solutions of how to select the conditioning pels are given, both of which are based on Rissanen's MDL criterion. Although no rigorous theory to account for the performance has been established, the experimental study demonstrates that one of them is superior in compression performance to any adaptive model utilizing a fixed number of conditioning pels. In this paper, an efficient implementation of models by means of an arithmetic code is also presented.
ER -