Synthetic aperture radar image segmentation by a detail preserving Markov random field approach

Author

Smits, Paul C. ; Dellepiane, Silvana G.

Author_Institution

Dept. of Biophys. & Electron. Eng., Genoa Univ., Italy

Volume

35

Issue

4

fYear

1997

fDate

7/1/1997 12:00:00 AM

Firstpage

844

Lastpage

857

Abstract

A multichannel image segmentation method is imposed that utilizes Markov random fields (MRFs) with adaptive neighborhood (AN) systems. Bayesian inference is applied to realize the combination of evidence from different knowledge sources. In such a way, optimization of the shape of a neighborhood system is achieved by following a criterion that makes use of the Markovian property exploiting the local image content. The MRF segmentation approach with AN systems (MRF-AN) makes it possible to better preserve small features and border areas. The purpose of the paper is to show the usefulness of the concept of MRF-AN for SAR image segmentation

Keywords

Bayes methods; Markov processes; adaptive signal processing; geophysical signal processing; geophysical techniques; image segmentation; radar imaging; remote sensing by radar; spaceborne radar; synthetic aperture radar; Bayes method; Bayesian inference; SAR; adaptive neighborhood system; adaptive signal processing; detail preserving Markov random field approach; geophysical measurement technique; image segmentation; knowledge source; land surface; multichannel image segmentation method; optimization; radar imaging; radar remote sensing; spaceborne radar; synthetic aperture radar; terrain mapping; Adaptive systems; Algorithm design and analysis; Bayesian methods; Image processing; Image segmentation; Markov random fields; Remote sensing; Roads; Shape; Synthetic aperture radar;

fLanguage

English

Journal_Title

Geoscience and Remote Sensing, IEEE Transactions on

Publisher

ieee

ISSN

0196-2892

Type

jour

DOI

10.1109/36.602527

Filename

602527