Specific land cover class mapping by semi-supervised weighted support vector machines

Research output: Contribution to journalArticle

8 Citations (Scopus)
16 Downloads (Pure)

Abstract

In many remote sensing projects on land cover mapping, the interest is often in a sub-set of classes presented in the study area. Conventional multi-class classification may lead to a considerable training effort and to the underestimation of the classes of interest. On the other hand, one-class classifiers require much less training, but may overestimate the real extension of the class of interest. This paper illustrates the combined use of cost-sensitive and semi-supervised learning to overcome these difficulties. This method utilises a manually-collected set of pixels of the class of interest and a random sample of pixels, keeping the training effort low. Each data point is then weighted according to its distance to its near positive data point to inform the learning algorithm. The proposed approach was compared with a conventional multi-class classifier, a one-class classifier, and a semi-supervised classifier in the discrimination of high-mangrove in Saloum estuary, Senegal, from Landsat imagery. The derived classification accuracies were high: 93.90% for the multi-class supervised classifier, 90.75% for the semi-supervised classifier, 88.75% for the one-class classifier, and 93.75% for the proposed method. The results show that accuracy achieved with the proposed method is statistically non-inferior to that achieved with standard binary classification, requiring however much less training effort.

Original languageEnglish
Article number181
JournalRemote Sensing
Volume9
Issue number2
DOIs
Publication statusPublished - 2017

Fingerprint

land cover
pixel
mangrove
Landsat
imagery
learning
estuary
remote sensing
support vector machine
cost
method

Keywords

  • Land cover
  • Landsat
  • Mangrove
  • One-class support vector machines
  • Random training set
  • Specific class mapping
  • Weighted support vector machine

Cite this

@article{1bcccd048c5a40fdb293d45f6e28c2c1,
title = "Specific land cover class mapping by semi-supervised weighted support vector machines",
abstract = "In many remote sensing projects on land cover mapping, the interest is often in a sub-set of classes presented in the study area. Conventional multi-class classification may lead to a considerable training effort and to the underestimation of the classes of interest. On the other hand, one-class classifiers require much less training, but may overestimate the real extension of the class of interest. This paper illustrates the combined use of cost-sensitive and semi-supervised learning to overcome these difficulties. This method utilises a manually-collected set of pixels of the class of interest and a random sample of pixels, keeping the training effort low. Each data point is then weighted according to its distance to its near positive data point to inform the learning algorithm. The proposed approach was compared with a conventional multi-class classifier, a one-class classifier, and a semi-supervised classifier in the discrimination of high-mangrove in Saloum estuary, Senegal, from Landsat imagery. The derived classification accuracies were high: 93.90{\%} for the multi-class supervised classifier, 90.75{\%} for the semi-supervised classifier, 88.75{\%} for the one-class classifier, and 93.75{\%} for the proposed method. The results show that accuracy achieved with the proposed method is statistically non-inferior to that achieved with standard binary classification, requiring however much less training effort.",
keywords = "Land cover, Landsat, Mangrove, One-class support vector machines, Random training set, Specific class mapping, Weighted support vector machine",
author = "Joel Silva and Fernando Ba{\cc}{\~a}o and Mario Caetano",
note = "Silva, J., Ba{\cc}{\~a}o, F., & Caetano, M. (2017). Specific land cover class mapping by semi-supervised weighted support vector machines. Remote Sensing, 9(2), [181]. https://doi.org/10.3390/rs9020181",
year = "2017",
doi = "10.3390/rs9020181",
language = "English",
volume = "9",
journal = "Remote Sensing",
issn = "2072-4292",
publisher = "Molecular Diversity Preservation International (MDPI)",
number = "2",

}

TY - JOUR

T1 - Specific land cover class mapping by semi-supervised weighted support vector machines

AU - Silva, Joel

AU - Bação, Fernando

AU - Caetano, Mario

N1 - Silva, J., Bação, F., & Caetano, M. (2017). Specific land cover class mapping by semi-supervised weighted support vector machines. Remote Sensing, 9(2), [181]. https://doi.org/10.3390/rs9020181

PY - 2017

Y1 - 2017

N2 - In many remote sensing projects on land cover mapping, the interest is often in a sub-set of classes presented in the study area. Conventional multi-class classification may lead to a considerable training effort and to the underestimation of the classes of interest. On the other hand, one-class classifiers require much less training, but may overestimate the real extension of the class of interest. This paper illustrates the combined use of cost-sensitive and semi-supervised learning to overcome these difficulties. This method utilises a manually-collected set of pixels of the class of interest and a random sample of pixels, keeping the training effort low. Each data point is then weighted according to its distance to its near positive data point to inform the learning algorithm. The proposed approach was compared with a conventional multi-class classifier, a one-class classifier, and a semi-supervised classifier in the discrimination of high-mangrove in Saloum estuary, Senegal, from Landsat imagery. The derived classification accuracies were high: 93.90% for the multi-class supervised classifier, 90.75% for the semi-supervised classifier, 88.75% for the one-class classifier, and 93.75% for the proposed method. The results show that accuracy achieved with the proposed method is statistically non-inferior to that achieved with standard binary classification, requiring however much less training effort.

AB - In many remote sensing projects on land cover mapping, the interest is often in a sub-set of classes presented in the study area. Conventional multi-class classification may lead to a considerable training effort and to the underestimation of the classes of interest. On the other hand, one-class classifiers require much less training, but may overestimate the real extension of the class of interest. This paper illustrates the combined use of cost-sensitive and semi-supervised learning to overcome these difficulties. This method utilises a manually-collected set of pixels of the class of interest and a random sample of pixels, keeping the training effort low. Each data point is then weighted according to its distance to its near positive data point to inform the learning algorithm. The proposed approach was compared with a conventional multi-class classifier, a one-class classifier, and a semi-supervised classifier in the discrimination of high-mangrove in Saloum estuary, Senegal, from Landsat imagery. The derived classification accuracies were high: 93.90% for the multi-class supervised classifier, 90.75% for the semi-supervised classifier, 88.75% for the one-class classifier, and 93.75% for the proposed method. The results show that accuracy achieved with the proposed method is statistically non-inferior to that achieved with standard binary classification, requiring however much less training effort.

KW - Land cover

KW - Landsat

KW - Mangrove

KW - One-class support vector machines

KW - Random training set

KW - Specific class mapping

KW - Weighted support vector machine

UR - http://www.scopus.com/inward/record.url?scp=85013625662&partnerID=8YFLogxK

U2 - 10.3390/rs9020181

DO - 10.3390/rs9020181

M3 - Article

VL - 9

JO - Remote Sensing

JF - Remote Sensing

SN - 2072-4292

IS - 2

M1 - 181

ER -