Adaptive box filters for removal of random noise from digital images

Eric M. Eliason, Alfred S. McEwen

Research output: Contribution to journalArticle

79 Citations (Scopus)

Abstract

We have developed adaptive box-filtering algorithms to (1) remove random bit errors (pixel values with no relation to the image scene) and (2) smooth noisy data (pixels related to the image scene but with an additive or multiplicative component of noise). For both procedures, we use the standard deviation (σ) of those pixels within a local box surrounding each pixel, hence they are adaptive filters. A series of two or three filters with decreasing box sizes can be run to clean up extremely noisy images and to remove bit errors near sharp edges. Our second filter, for noise smoothing, is identical to the 'sigma filter' of Lee (1983a) except that we use the local (adaptive) σ rather than a fixed σ. The filter averages only those pixels within the box that have intensities within 1.0 to 2.0 σ of the central pixel. This technique effectively reduces speckle in radar images without eliminating fine details.

Original languageEnglish (US)
Pages (from-to)453-458
Number of pages6
JournalPhotogrammetric Engineering and Remote Sensing
Volume56
Issue number4
StatePublished - Apr 1990
Externally publishedYes

Fingerprint

digital image
pixel
Pixels
filter
speckle
Adaptive filters
Speckle
smoothing
removal
Radar
radar

ASJC Scopus subject areas

  • Earth and Planetary Sciences(all)
  • Environmental Science(all)

Cite this

Adaptive box filters for removal of random noise from digital images. / Eliason, Eric M.; McEwen, Alfred S.

In: Photogrammetric Engineering and Remote Sensing, Vol. 56, No. 4, 04.1990, p. 453-458.

Research output: Contribution to journalArticle

@article{516c6e87ffd14d38aa008b959af08a94,
title = "Adaptive box filters for removal of random noise from digital images",
abstract = "We have developed adaptive box-filtering algorithms to (1) remove random bit errors (pixel values with no relation to the image scene) and (2) smooth noisy data (pixels related to the image scene but with an additive or multiplicative component of noise). For both procedures, we use the standard deviation (σ) of those pixels within a local box surrounding each pixel, hence they are adaptive filters. A series of two or three filters with decreasing box sizes can be run to clean up extremely noisy images and to remove bit errors near sharp edges. Our second filter, for noise smoothing, is identical to the 'sigma filter' of Lee (1983a) except that we use the local (adaptive) σ rather than a fixed σ. The filter averages only those pixels within the box that have intensities within 1.0 to 2.0 σ of the central pixel. This technique effectively reduces speckle in radar images without eliminating fine details.",
author = "Eliason, {Eric M.} and McEwen, {Alfred S.}",
year = "1990",
month = "4",
language = "English (US)",
volume = "56",
pages = "453--458",
journal = "Photogrammetric Engineering and Remote Sensing",
issn = "0099-1112",
publisher = "American Society for Photogrammetry and Remote Sensing",
number = "4",

}

TY - JOUR

T1 - Adaptive box filters for removal of random noise from digital images

AU - Eliason, Eric M.

AU - McEwen, Alfred S.

PY - 1990/4

Y1 - 1990/4

N2 - We have developed adaptive box-filtering algorithms to (1) remove random bit errors (pixel values with no relation to the image scene) and (2) smooth noisy data (pixels related to the image scene but with an additive or multiplicative component of noise). For both procedures, we use the standard deviation (σ) of those pixels within a local box surrounding each pixel, hence they are adaptive filters. A series of two or three filters with decreasing box sizes can be run to clean up extremely noisy images and to remove bit errors near sharp edges. Our second filter, for noise smoothing, is identical to the 'sigma filter' of Lee (1983a) except that we use the local (adaptive) σ rather than a fixed σ. The filter averages only those pixels within the box that have intensities within 1.0 to 2.0 σ of the central pixel. This technique effectively reduces speckle in radar images without eliminating fine details.

AB - We have developed adaptive box-filtering algorithms to (1) remove random bit errors (pixel values with no relation to the image scene) and (2) smooth noisy data (pixels related to the image scene but with an additive or multiplicative component of noise). For both procedures, we use the standard deviation (σ) of those pixels within a local box surrounding each pixel, hence they are adaptive filters. A series of two or three filters with decreasing box sizes can be run to clean up extremely noisy images and to remove bit errors near sharp edges. Our second filter, for noise smoothing, is identical to the 'sigma filter' of Lee (1983a) except that we use the local (adaptive) σ rather than a fixed σ. The filter averages only those pixels within the box that have intensities within 1.0 to 2.0 σ of the central pixel. This technique effectively reduces speckle in radar images without eliminating fine details.

UR - http://www.scopus.com/inward/record.url?scp=0025573208&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0025573208&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:0025573208

VL - 56

SP - 453

EP - 458

JO - Photogrammetric Engineering and Remote Sensing

JF - Photogrammetric Engineering and Remote Sensing

SN - 0099-1112

IS - 4

ER -