Multiscale superpixel segmentation-based band expansion for change detection

ERTÜRK A.

REMOTE SENSING LETTERS, vol.14, no.5, pp.534-544, 2023 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 14 Issue: 5
  • Publication Date: 2023
  • Doi Number: 10.1080/2150704x.2023.2217527
  • Journal Name: REMOTE SENSING LETTERS
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, PASCAL, Aerospace Database, Agricultural & Environmental Science Database, Applied Science & Technology Source, BIOSIS, CAB Abstracts, Compendex, Computer & Applied Sciences, Environment Index, Geobase, INSPEC, Civil Engineering Abstracts
  • Page Numbers: pp.534-544
  • Kocaeli University Affiliated: Yes

Abstract

Change detection (CD) for remotely sensed images has gained great relevance in the last decade due to an increase in the number of Earth Observation (EO) missions, improved temporal resolutions, and open data policies. However, efficient exploitation and integration of spatial information for CD remains a challenging issue. This paper proposes a multiscale superpixel segmentation-based band expansion approach to address these issues. Superpixels derived at different scales are used to extract spatial information of objects and landscapes of various sizes, and this information is integrated in the form of new spectral bands. The expanded band set may then be used with any CD method, whether for binary or multi-class CD. Experimental results on multispectral (MS) and hyperspectral (HS) multitemporal images validate the proposed approach's performance benefit.