Post classification smoothing in sub-decimeter resolution images with semi-supervised label propagation

John E. Vargas-Munoz, Devis Tuia, Jefersson A. Dos Santos, Alexandre X. Falcao

Research output: Chapter in Book/Report/Conference proceedingConference paperAcademicpeer-review

Abstract

In this paper, we propose a post classification smoothing method aimed at improving the accuracy and visual appearance of sub-decimeter image classification results. Starting from the class confidence maps of a supervised classifier, we find a set of high confidence markers and propagate labels on an extended region adjacency graph. We apply the proposed method on a challenging 5cm resolution dataset over Potsdam, Germany. The proposed algorithm outperforms state-of-the-art post classification smoothing algorithms both when the classifier is trained specifically on the image and when it is trained and tested in different set of images.

Original languageEnglish
Title of host publication2017 IEEE International Geoscience and Remote Sensing Symposium
Subtitle of host publicationInternational Cooperation for Global Awareness, IGARSS 2017 - Proceedings
Place of PublicationFort Worth
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages3688-3691
Number of pages4
ISBN (Electronic)9781509049516
ISBN (Print)9781509049523
DOIs
Publication statusPublished - 1 Dec 2017
Externally publishedYes
Event37th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2017 - Fort Worth, United States
Duration: 23 Jul 201728 Jul 2017

Publication series

NameInternational Geoscience and Remote Sensing Symposium (IGARSS)
Volume2017-July

Conference

Conference37th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2017
CountryUnited States
CityFort Worth
Period23/07/1728/07/17

Keywords

  • Contextual classification
  • Image foresting transform
  • Markov random fields
  • Superpixel

Fingerprint Dive into the research topics of 'Post classification smoothing in sub-decimeter resolution images with semi-supervised label propagation'. Together they form a unique fingerprint.

Cite this