Remote sensing image analysis by aggregation of segmentation-classification collaborative agents

Andrés Troya-Galvis, Pierre Gançarski, Laure Berti-Equille

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

In this article we present two different approaches for automatic remote sensing image interpretation which are based on a multi-paradigm collaborative framework which uses classification in order to guide the segmentation process. The first approach applies sequentially many one-vs-all class extractors in a manner inspired by cascading techniques in machine learning. The second approach applies many collaborating one-vs-all class extractors in parallel. We show that the collaboration of the segmentation and classification paradigms result in a remarkable reduction of segmentation errors but also in better object classification in comparison to a hybrid pixel-object approach as well as a deep learning approach.

Original languageEnglish
Pages (from-to)259-274
Number of pages16
JournalPattern Recognition
Volume73
DOIs
Publication statusPublished - 1 Jan 2018

Fingerprint

Image analysis
Remote sensing
Agglomeration
Learning systems
Pixels
Deep learning

Keywords

  • Classification
  • Collaborative approaches
  • Remote sensing
  • Segmentation

ASJC Scopus subject areas

  • Software
  • Signal Processing
  • Computer Vision and Pattern Recognition
  • Artificial Intelligence

Cite this

Remote sensing image analysis by aggregation of segmentation-classification collaborative agents. / Troya-Galvis, Andrés; Gançarski, Pierre; Berti-Equille, Laure.

In: Pattern Recognition, Vol. 73, 01.01.2018, p. 259-274.

Research output: Contribution to journalArticle

@article{e58f1ca5a0b2465dabf57b00122dbef0,
title = "Remote sensing image analysis by aggregation of segmentation-classification collaborative agents",
abstract = "In this article we present two different approaches for automatic remote sensing image interpretation which are based on a multi-paradigm collaborative framework which uses classification in order to guide the segmentation process. The first approach applies sequentially many one-vs-all class extractors in a manner inspired by cascading techniques in machine learning. The second approach applies many collaborating one-vs-all class extractors in parallel. We show that the collaboration of the segmentation and classification paradigms result in a remarkable reduction of segmentation errors but also in better object classification in comparison to a hybrid pixel-object approach as well as a deep learning approach.",
keywords = "Classification, Collaborative approaches, Remote sensing, Segmentation",
author = "Andr{\'e}s Troya-Galvis and Pierre Gan{\cc}arski and Laure Berti-Equille",
year = "2018",
month = "1",
day = "1",
doi = "10.1016/j.patcog.2017.08.030",
language = "English",
volume = "73",
pages = "259--274",
journal = "Pattern Recognition",
issn = "0031-3203",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - Remote sensing image analysis by aggregation of segmentation-classification collaborative agents

AU - Troya-Galvis, Andrés

AU - Gançarski, Pierre

AU - Berti-Equille, Laure

PY - 2018/1/1

Y1 - 2018/1/1

N2 - In this article we present two different approaches for automatic remote sensing image interpretation which are based on a multi-paradigm collaborative framework which uses classification in order to guide the segmentation process. The first approach applies sequentially many one-vs-all class extractors in a manner inspired by cascading techniques in machine learning. The second approach applies many collaborating one-vs-all class extractors in parallel. We show that the collaboration of the segmentation and classification paradigms result in a remarkable reduction of segmentation errors but also in better object classification in comparison to a hybrid pixel-object approach as well as a deep learning approach.

AB - In this article we present two different approaches for automatic remote sensing image interpretation which are based on a multi-paradigm collaborative framework which uses classification in order to guide the segmentation process. The first approach applies sequentially many one-vs-all class extractors in a manner inspired by cascading techniques in machine learning. The second approach applies many collaborating one-vs-all class extractors in parallel. We show that the collaboration of the segmentation and classification paradigms result in a remarkable reduction of segmentation errors but also in better object classification in comparison to a hybrid pixel-object approach as well as a deep learning approach.

KW - Classification

KW - Collaborative approaches

KW - Remote sensing

KW - Segmentation

UR - http://www.scopus.com/inward/record.url?scp=85028728912&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85028728912&partnerID=8YFLogxK

U2 - 10.1016/j.patcog.2017.08.030

DO - 10.1016/j.patcog.2017.08.030

M3 - Article

AN - SCOPUS:85028728912

VL - 73

SP - 259

EP - 274

JO - Pattern Recognition

JF - Pattern Recognition

SN - 0031-3203

ER -