NATURAL PIXEL DECOMPOSITION FOR TWO-DIMENSIONAL IMAGE RECONSTRUCTION.

Michael H. Buonocore, William R. Brody, Albert Macovski

Research output: Contribution to journalArticle

88 Citations (Scopus)

Abstract

In two-dimensional image reconstruction from line integrals using maximum likelihood, Bayesian, or minimum variance algorithms, the x-y plane on which the object estimate is defined is decomposed into nonoverlapping regions, or ″pixels″ . This decompostion of an otherwise continuous structure results in significant errors, or model noise, which can exceed the effects of the fundamental measurement noise. Many applications of diagnostic cross-sectional imaging require that images be obtained from limited data. A new formalism provides a connection between the continuous object to be reconstructed and its discrete representation. Using this formalism, the authors describe a decomposition of the x-y plane into a set of discrete, but overlapping pixels. In this model the pixels are uniquely defined by the beam paths. These ″natural″ pixels provide new, discrete filtered back projection reconstruction algorithms for limited data. The resulting reconstruction system not only avoids the pixel partitioning error, but provides basic mathematical properties which lead to profound computational advantages.

Original languageEnglish (US)
Pages (from-to)69-78
Number of pages10
JournalIEEE Transactions on Biomedical Engineering
VolumeBME-28
Issue number2
StatePublished - Feb 1981
Externally publishedYes

Fingerprint

Image reconstruction
Pixels
Decomposition
Maximum likelihood
Imaging techniques

ASJC Scopus subject areas

  • Biomedical Engineering

Cite this

Buonocore, M. H., Brody, W. R., & Macovski, A. (1981). NATURAL PIXEL DECOMPOSITION FOR TWO-DIMENSIONAL IMAGE RECONSTRUCTION. IEEE Transactions on Biomedical Engineering, BME-28(2), 69-78.

NATURAL PIXEL DECOMPOSITION FOR TWO-DIMENSIONAL IMAGE RECONSTRUCTION. / Buonocore, Michael H.; Brody, William R.; Macovski, Albert.

In: IEEE Transactions on Biomedical Engineering, Vol. BME-28, No. 2, 02.1981, p. 69-78.

Research output: Contribution to journalArticle

Buonocore, MH, Brody, WR & Macovski, A 1981, 'NATURAL PIXEL DECOMPOSITION FOR TWO-DIMENSIONAL IMAGE RECONSTRUCTION.', IEEE Transactions on Biomedical Engineering, vol. BME-28, no. 2, pp. 69-78.
Buonocore, Michael H. ; Brody, William R. ; Macovski, Albert. / NATURAL PIXEL DECOMPOSITION FOR TWO-DIMENSIONAL IMAGE RECONSTRUCTION. In: IEEE Transactions on Biomedical Engineering. 1981 ; Vol. BME-28, No. 2. pp. 69-78.
@article{f9954228b2b84e9b897d1c331a60b125,
title = "NATURAL PIXEL DECOMPOSITION FOR TWO-DIMENSIONAL IMAGE RECONSTRUCTION.",
abstract = "In two-dimensional image reconstruction from line integrals using maximum likelihood, Bayesian, or minimum variance algorithms, the x-y plane on which the object estimate is defined is decomposed into nonoverlapping regions, or ″pixels″ . This decompostion of an otherwise continuous structure results in significant errors, or model noise, which can exceed the effects of the fundamental measurement noise. Many applications of diagnostic cross-sectional imaging require that images be obtained from limited data. A new formalism provides a connection between the continuous object to be reconstructed and its discrete representation. Using this formalism, the authors describe a decomposition of the x-y plane into a set of discrete, but overlapping pixels. In this model the pixels are uniquely defined by the beam paths. These ″natural″ pixels provide new, discrete filtered back projection reconstruction algorithms for limited data. The resulting reconstruction system not only avoids the pixel partitioning error, but provides basic mathematical properties which lead to profound computational advantages.",
author = "Buonocore, {Michael H.} and Brody, {William R.} and Albert Macovski",
year = "1981",
month = "2",
language = "English (US)",
volume = "BME-28",
pages = "69--78",
journal = "IEEE Transactions on Biomedical Engineering",
issn = "0018-9294",
publisher = "IEEE Computer Society",
number = "2",

}

TY - JOUR

T1 - NATURAL PIXEL DECOMPOSITION FOR TWO-DIMENSIONAL IMAGE RECONSTRUCTION.

AU - Buonocore, Michael H.

AU - Brody, William R.

AU - Macovski, Albert

PY - 1981/2

Y1 - 1981/2

N2 - In two-dimensional image reconstruction from line integrals using maximum likelihood, Bayesian, or minimum variance algorithms, the x-y plane on which the object estimate is defined is decomposed into nonoverlapping regions, or ″pixels″ . This decompostion of an otherwise continuous structure results in significant errors, or model noise, which can exceed the effects of the fundamental measurement noise. Many applications of diagnostic cross-sectional imaging require that images be obtained from limited data. A new formalism provides a connection between the continuous object to be reconstructed and its discrete representation. Using this formalism, the authors describe a decomposition of the x-y plane into a set of discrete, but overlapping pixels. In this model the pixels are uniquely defined by the beam paths. These ″natural″ pixels provide new, discrete filtered back projection reconstruction algorithms for limited data. The resulting reconstruction system not only avoids the pixel partitioning error, but provides basic mathematical properties which lead to profound computational advantages.

AB - In two-dimensional image reconstruction from line integrals using maximum likelihood, Bayesian, or minimum variance algorithms, the x-y plane on which the object estimate is defined is decomposed into nonoverlapping regions, or ″pixels″ . This decompostion of an otherwise continuous structure results in significant errors, or model noise, which can exceed the effects of the fundamental measurement noise. Many applications of diagnostic cross-sectional imaging require that images be obtained from limited data. A new formalism provides a connection between the continuous object to be reconstructed and its discrete representation. Using this formalism, the authors describe a decomposition of the x-y plane into a set of discrete, but overlapping pixels. In this model the pixels are uniquely defined by the beam paths. These ″natural″ pixels provide new, discrete filtered back projection reconstruction algorithms for limited data. The resulting reconstruction system not only avoids the pixel partitioning error, but provides basic mathematical properties which lead to profound computational advantages.

UR - http://www.scopus.com/inward/record.url?scp=0019394026&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0019394026&partnerID=8YFLogxK

M3 - Article

VL - BME-28

SP - 69

EP - 78

JO - IEEE Transactions on Biomedical Engineering

JF - IEEE Transactions on Biomedical Engineering

SN - 0018-9294

IS - 2

ER -