Unsupervised classification of multilook polarimetric SAR data using spatially variant wishart mixture model with double constraints

Chi Liu, Wenzhi Liao, Heng-Chao Li, Kun Fu, Wilfried Philips

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

This paper addresses the unsupervised classification problems for multilook Polarimetric synthetic aperture radar (PolSAR) images by proposing a patch-level spatially variant Wishart mixture model (SVWMM) with double constraints. We construct this model by jointly modeling the pixels in a patch (rather than an individual pixel) so as to effectively capture the local correlation in the PolSAR images. More importantly, a responsibility parameter is introduced to the proposed model, providing not only the possibility to represent the importance of different pixels within a patch but also the additional flexibility for incorporating the spatial information. As such, double constraints are further imposed by simultaneously utilizing the similarities of the neighboring pixels, respectively, defined on two different parameter spaces (i.e., the hyperparameter in the posterior distribution of mixing coefficients and the responsibility parameter). Furthermore, the variational inference algorithm is developed to achieve effective learning of the proposed SVWMM with the closed-form updates, facilitating the automatic determination of the cluster number. Experimental results on several PolSAR data sets from both airborne and spaceborne sensors demonstrate that the proposed method is effective and it enables better performances on unsupervised classification than the conventional methods.
LanguageEnglish
Pages5600-5613
Number of pages14
JournalIEEE Transactions on Geoscience and Remote Sensing
Volume56
Issue number10
DOIs
Publication statusPublished - 23 Apr 2018

Fingerprint

unsupervised classification
pixel
synthetic aperture radar
Synthetic aperture radar
Pixels
learning
sensor
Sensors
modeling
parameter
responsibility
method

Keywords

  • polarimetric synthetic aperture radar (PolSAR)
  • spatial constraint
  • unsupervised classification
  • variational inference (VI)
  • wishart mixture model (WMM)

Cite this

@article{056b77bde9774e9d9c3f8548b9f2c45b,
title = "Unsupervised classification of multilook polarimetric SAR data using spatially variant wishart mixture model with double constraints",
abstract = "This paper addresses the unsupervised classification problems for multilook Polarimetric synthetic aperture radar (PolSAR) images by proposing a patch-level spatially variant Wishart mixture model (SVWMM) with double constraints. We construct this model by jointly modeling the pixels in a patch (rather than an individual pixel) so as to effectively capture the local correlation in the PolSAR images. More importantly, a responsibility parameter is introduced to the proposed model, providing not only the possibility to represent the importance of different pixels within a patch but also the additional flexibility for incorporating the spatial information. As such, double constraints are further imposed by simultaneously utilizing the similarities of the neighboring pixels, respectively, defined on two different parameter spaces (i.e., the hyperparameter in the posterior distribution of mixing coefficients and the responsibility parameter). Furthermore, the variational inference algorithm is developed to achieve effective learning of the proposed SVWMM with the closed-form updates, facilitating the automatic determination of the cluster number. Experimental results on several PolSAR data sets from both airborne and spaceborne sensors demonstrate that the proposed method is effective and it enables better performances on unsupervised classification than the conventional methods.",
keywords = "polarimetric synthetic aperture radar (PolSAR), spatial constraint, unsupervised classification, variational inference (VI), wishart mixture model (WMM)",
author = "Chi Liu and Wenzhi Liao and Heng-Chao Li and Kun Fu and Wilfried Philips",
year = "2018",
month = "4",
day = "23",
doi = "10.1109/tgrs.2018.2819995",
language = "English",
volume = "56",
pages = "5600--5613",
journal = "IEEE Transactions on Geoscience and Remote Sensing",
issn = "0196-2892",
publisher = "IEEE",
number = "10",

}

Unsupervised classification of multilook polarimetric SAR data using spatially variant wishart mixture model with double constraints. / Liu, Chi; Liao, Wenzhi; Li, Heng-Chao; Fu, Kun; Philips, Wilfried.

In: IEEE Transactions on Geoscience and Remote Sensing, Vol. 56, No. 10, 23.04.2018, p. 5600-5613.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Unsupervised classification of multilook polarimetric SAR data using spatially variant wishart mixture model with double constraints

AU - Liu, Chi

AU - Liao, Wenzhi

AU - Li, Heng-Chao

AU - Fu, Kun

AU - Philips, Wilfried

PY - 2018/4/23

Y1 - 2018/4/23

N2 - This paper addresses the unsupervised classification problems for multilook Polarimetric synthetic aperture radar (PolSAR) images by proposing a patch-level spatially variant Wishart mixture model (SVWMM) with double constraints. We construct this model by jointly modeling the pixels in a patch (rather than an individual pixel) so as to effectively capture the local correlation in the PolSAR images. More importantly, a responsibility parameter is introduced to the proposed model, providing not only the possibility to represent the importance of different pixels within a patch but also the additional flexibility for incorporating the spatial information. As such, double constraints are further imposed by simultaneously utilizing the similarities of the neighboring pixels, respectively, defined on two different parameter spaces (i.e., the hyperparameter in the posterior distribution of mixing coefficients and the responsibility parameter). Furthermore, the variational inference algorithm is developed to achieve effective learning of the proposed SVWMM with the closed-form updates, facilitating the automatic determination of the cluster number. Experimental results on several PolSAR data sets from both airborne and spaceborne sensors demonstrate that the proposed method is effective and it enables better performances on unsupervised classification than the conventional methods.

AB - This paper addresses the unsupervised classification problems for multilook Polarimetric synthetic aperture radar (PolSAR) images by proposing a patch-level spatially variant Wishart mixture model (SVWMM) with double constraints. We construct this model by jointly modeling the pixels in a patch (rather than an individual pixel) so as to effectively capture the local correlation in the PolSAR images. More importantly, a responsibility parameter is introduced to the proposed model, providing not only the possibility to represent the importance of different pixels within a patch but also the additional flexibility for incorporating the spatial information. As such, double constraints are further imposed by simultaneously utilizing the similarities of the neighboring pixels, respectively, defined on two different parameter spaces (i.e., the hyperparameter in the posterior distribution of mixing coefficients and the responsibility parameter). Furthermore, the variational inference algorithm is developed to achieve effective learning of the proposed SVWMM with the closed-form updates, facilitating the automatic determination of the cluster number. Experimental results on several PolSAR data sets from both airborne and spaceborne sensors demonstrate that the proposed method is effective and it enables better performances on unsupervised classification than the conventional methods.

KW - polarimetric synthetic aperture radar (PolSAR)

KW - spatial constraint

KW - unsupervised classification

KW - variational inference (VI)

KW - wishart mixture model (WMM)

UR - http://hdl.handle.net/1854/LU-8559413

U2 - 10.1109/tgrs.2018.2819995

DO - 10.1109/tgrs.2018.2819995

M3 - Article

VL - 56

SP - 5600

EP - 5613

JO - IEEE Transactions on Geoscience and Remote Sensing

T2 - IEEE Transactions on Geoscience and Remote Sensing

JF - IEEE Transactions on Geoscience and Remote Sensing

SN - 0196-2892

IS - 10

ER -