Unsupervised exploration of hyperspectral and multispectral images

Research output: Chapter in Book/Report/Conference proceedingBook chapterResearchpeer-review

  • Federico Marini
  • José Manuel Amigo

One of the first actions to make in the analysis of hyperspectral and multispectral images is the unsupervised exploration of the spatio-spectral domains. Unsupervised exploration techniques are methods that obtain information about the spatial distribution of compounds on the images, some of their spectral signatures, their main sources of variation, and also help to detect defectuous pixels or spectra, by only using the spatial and spectral information of the images acquired in an unsupervised manner. In this chapter, we present the most popular methods for unsupervised modeling together with examples to understand their major benefits and drawbacks.

Original languageEnglish
Title of host publicationHyperspectral Imaging
EditorsJosé Manuel Amigo
Number of pages22
PublisherElsevier
Publication date2020
Pages93-114
Chapter2.4
ISBN (Print)978-0-444-63977-6
DOIs
Publication statusPublished - 2020
SeriesData Handling in Science and Technology
Volume32
ISSN0922-3487

    Research areas

  • Clusters, Dendrograms, Fuzzy clustering, K-means, Multivariate data analysis, PCA, Unsupervised

ID: 231241111