A new fully convolutional neural network for semantic segmentation of polarimetric SAR imagery in complex land cover ecosystem

Fariba Mohammadimanesh (Corresponding Author), Bahram Salehi, Masoud Mahdianpari, Eric Gill, Matthieu Molinier

Research output: Contribution to journalArticleScientificpeer-review

1 Citation (Scopus)

Abstract

Despite the application of state-of-the-art fully Convolutional Neural Networks (CNNs) for semantic segmentation of very high-resolution optical imagery, their capacity has not yet been thoroughly examined for the classification of Synthetic Aperture Radar (SAR) images. The presence of speckle noise, the absence of efficient feature expression, and the limited availability of labelled SAR samples have hindered the application of the state-of-the-art CNNs for the classification of SAR imagery. This is of great concern for mapping complex land cover ecosystems, such as wetlands, where backscattering/spectrally similar signatures of land cover units further complicate the matter. Accordingly, we propose a new Fully Convolutional Network (FCN) architecture that can be trained in an end-to-end scheme and is specifically designed for the classification of wetland complexes using polarimetric SAR (PolSAR) imagery. The proposed architecture follows an encoder-decoder paradigm, wherein the input data are fed into a stack of convolutional filters (encoder) to extract high-level abstract features and a stack of transposed convolutional filters (decoder) to gradually up-sample the low resolution output to the spatial resolution of the original input image. The proposed network also benefits from recent advances in CNN designs, namely the addition of inception modules and skip connections with residual units. The former component improves multi-scale inference and enriches contextual information, while the latter contributes to the recovery of more detailed information and simplifies optimization. Moreover, an in-depth investigation of the learned features via opening the black box demonstrates that convolutional filters extract discriminative polarimetric features, thus mitigating the limitation of the feature engineering design in PolSAR image processing. Experimental results from full polarimetric RADARSAT-2 imagery illustrate that the proposed network outperforms the conventional random forest classifier and the state-of-the-art FCNs, such as FCN-32s, FCN-16s, FCN-8s, and SegNet, both visually and numerically for wetland mapping.

Original languageEnglish
Pages (from-to)223-236
Number of pages14
JournalISPRS Journal of Photogrammetry and Remote Sensing
Volume151
DOIs
Publication statusPublished - 2019
MoE publication typeNot Eligible

Fingerprint

radar imagery
semantics
ecosystems
synthetic aperture radar
Synthetic aperture radar
Ecosystems
segmentation
land cover
Wetlands
Semantics
Neural networks
ecosystem
wetlands
wetland
filter
imagery
Backscattering
Speckle
decoders
Network architecture

Keywords

  • Convolutional Neural Network (CNN)
  • Deep learning
  • Encoder-decoder
  • Fully Convolutional Network (FCN)
  • Land cover
  • Polarimetric Synthetic Aperture Radar (PolSAR)
  • Wetland

Cite this

@article{42e60c5ee05e480e8b1b01291024c7b9,
title = "A new fully convolutional neural network for semantic segmentation of polarimetric SAR imagery in complex land cover ecosystem",
abstract = "Despite the application of state-of-the-art fully Convolutional Neural Networks (CNNs) for semantic segmentation of very high-resolution optical imagery, their capacity has not yet been thoroughly examined for the classification of Synthetic Aperture Radar (SAR) images. The presence of speckle noise, the absence of efficient feature expression, and the limited availability of labelled SAR samples have hindered the application of the state-of-the-art CNNs for the classification of SAR imagery. This is of great concern for mapping complex land cover ecosystems, such as wetlands, where backscattering/spectrally similar signatures of land cover units further complicate the matter. Accordingly, we propose a new Fully Convolutional Network (FCN) architecture that can be trained in an end-to-end scheme and is specifically designed for the classification of wetland complexes using polarimetric SAR (PolSAR) imagery. The proposed architecture follows an encoder-decoder paradigm, wherein the input data are fed into a stack of convolutional filters (encoder) to extract high-level abstract features and a stack of transposed convolutional filters (decoder) to gradually up-sample the low resolution output to the spatial resolution of the original input image. The proposed network also benefits from recent advances in CNN designs, namely the addition of inception modules and skip connections with residual units. The former component improves multi-scale inference and enriches contextual information, while the latter contributes to the recovery of more detailed information and simplifies optimization. Moreover, an in-depth investigation of the learned features via opening the black box demonstrates that convolutional filters extract discriminative polarimetric features, thus mitigating the limitation of the feature engineering design in PolSAR image processing. Experimental results from full polarimetric RADARSAT-2 imagery illustrate that the proposed network outperforms the conventional random forest classifier and the state-of-the-art FCNs, such as FCN-32s, FCN-16s, FCN-8s, and SegNet, both visually and numerically for wetland mapping.",
keywords = "Convolutional Neural Network (CNN), Deep learning, Encoder-decoder, Fully Convolutional Network (FCN), Land cover, Polarimetric Synthetic Aperture Radar (PolSAR), Wetland",
author = "Fariba Mohammadimanesh and Bahram Salehi and Masoud Mahdianpari and Eric Gill and Matthieu Molinier",
year = "2019",
doi = "10.1016/j.isprsjprs.2019.03.015",
language = "English",
volume = "151",
pages = "223--236",
journal = "ISPRS Journal of Photogrammetry and Remote Sensing",
issn = "0924-2716",
publisher = "Elsevier",

}

A new fully convolutional neural network for semantic segmentation of polarimetric SAR imagery in complex land cover ecosystem. / Mohammadimanesh, Fariba (Corresponding Author); Salehi, Bahram; Mahdianpari, Masoud; Gill, Eric; Molinier, Matthieu.

In: ISPRS Journal of Photogrammetry and Remote Sensing, Vol. 151, 2019, p. 223-236.

Research output: Contribution to journalArticleScientificpeer-review

TY - JOUR

T1 - A new fully convolutional neural network for semantic segmentation of polarimetric SAR imagery in complex land cover ecosystem

AU - Mohammadimanesh, Fariba

AU - Salehi, Bahram

AU - Mahdianpari, Masoud

AU - Gill, Eric

AU - Molinier, Matthieu

PY - 2019

Y1 - 2019

N2 - Despite the application of state-of-the-art fully Convolutional Neural Networks (CNNs) for semantic segmentation of very high-resolution optical imagery, their capacity has not yet been thoroughly examined for the classification of Synthetic Aperture Radar (SAR) images. The presence of speckle noise, the absence of efficient feature expression, and the limited availability of labelled SAR samples have hindered the application of the state-of-the-art CNNs for the classification of SAR imagery. This is of great concern for mapping complex land cover ecosystems, such as wetlands, where backscattering/spectrally similar signatures of land cover units further complicate the matter. Accordingly, we propose a new Fully Convolutional Network (FCN) architecture that can be trained in an end-to-end scheme and is specifically designed for the classification of wetland complexes using polarimetric SAR (PolSAR) imagery. The proposed architecture follows an encoder-decoder paradigm, wherein the input data are fed into a stack of convolutional filters (encoder) to extract high-level abstract features and a stack of transposed convolutional filters (decoder) to gradually up-sample the low resolution output to the spatial resolution of the original input image. The proposed network also benefits from recent advances in CNN designs, namely the addition of inception modules and skip connections with residual units. The former component improves multi-scale inference and enriches contextual information, while the latter contributes to the recovery of more detailed information and simplifies optimization. Moreover, an in-depth investigation of the learned features via opening the black box demonstrates that convolutional filters extract discriminative polarimetric features, thus mitigating the limitation of the feature engineering design in PolSAR image processing. Experimental results from full polarimetric RADARSAT-2 imagery illustrate that the proposed network outperforms the conventional random forest classifier and the state-of-the-art FCNs, such as FCN-32s, FCN-16s, FCN-8s, and SegNet, both visually and numerically for wetland mapping.

AB - Despite the application of state-of-the-art fully Convolutional Neural Networks (CNNs) for semantic segmentation of very high-resolution optical imagery, their capacity has not yet been thoroughly examined for the classification of Synthetic Aperture Radar (SAR) images. The presence of speckle noise, the absence of efficient feature expression, and the limited availability of labelled SAR samples have hindered the application of the state-of-the-art CNNs for the classification of SAR imagery. This is of great concern for mapping complex land cover ecosystems, such as wetlands, where backscattering/spectrally similar signatures of land cover units further complicate the matter. Accordingly, we propose a new Fully Convolutional Network (FCN) architecture that can be trained in an end-to-end scheme and is specifically designed for the classification of wetland complexes using polarimetric SAR (PolSAR) imagery. The proposed architecture follows an encoder-decoder paradigm, wherein the input data are fed into a stack of convolutional filters (encoder) to extract high-level abstract features and a stack of transposed convolutional filters (decoder) to gradually up-sample the low resolution output to the spatial resolution of the original input image. The proposed network also benefits from recent advances in CNN designs, namely the addition of inception modules and skip connections with residual units. The former component improves multi-scale inference and enriches contextual information, while the latter contributes to the recovery of more detailed information and simplifies optimization. Moreover, an in-depth investigation of the learned features via opening the black box demonstrates that convolutional filters extract discriminative polarimetric features, thus mitigating the limitation of the feature engineering design in PolSAR image processing. Experimental results from full polarimetric RADARSAT-2 imagery illustrate that the proposed network outperforms the conventional random forest classifier and the state-of-the-art FCNs, such as FCN-32s, FCN-16s, FCN-8s, and SegNet, both visually and numerically for wetland mapping.

KW - Convolutional Neural Network (CNN)

KW - Deep learning

KW - Encoder-decoder

KW - Fully Convolutional Network (FCN)

KW - Land cover

KW - Polarimetric Synthetic Aperture Radar (PolSAR)

KW - Wetland

UR - http://www.scopus.com/inward/record.url?scp=85063541232&partnerID=8YFLogxK

U2 - 10.1016/j.isprsjprs.2019.03.015

DO - 10.1016/j.isprsjprs.2019.03.015

M3 - Article

VL - 151

SP - 223

EP - 236

JO - ISPRS Journal of Photogrammetry and Remote Sensing

JF - ISPRS Journal of Photogrammetry and Remote Sensing

SN - 0924-2716

ER -