@article{fdi:010084215,
  title = {{P}robabilistic unsupervised classification for large-scale analysis of spectral imaging data},
  author = {{P}aradis, {E}mmanuel},
  editor = {},
  language = {{ENG}},
  abstract = {{L}and cover classification of remote sensing data is a fundamental tool to study changes in the environment such as deforestation or wildfires. {A} current challenge is to quantify land cover changes with real-time, large-scale data from modern hyper- or multispectral sensors. {A} range of methods are available for this task, several of them being based on the k-means classification method which is efficient when classes of land cover are well separated. {H}ere a new algorithm, called probabilistic k-means, is presented to solve some of the limitations of the standard k-means. {I}t is shown that the new algorithm performs better than the standard k-means when the data are noisy. {I}f the number of land cover classes is unknown, an entropy-based criterion can be used to select the best number of classes. {T}he proposed new algorithm is implemented in a combination of {R} and {C} computer codes which is particularly efficient with large data sets: a whole image with more than 3 million pixels and covering more than 10,000 km2 can be analysed in a few minutes. {F}our applications with hyperspectral and multispectral data are presented. {F}or the data sets with ground truth data, the overall accuracy of the probabilistic k-means was substantially improved compared to the standard k-means. {O}ne of these data sets includes more than 120 million pixels, demonstrating the scalability of the proposed approach. {T}hese developments open new perspectives for the large scale analysis of remote sensing data. {A}ll computer code are available in an open-source package called sentinel.},
  keywords = {{U}nsupervised classification ; k-means ; {L}and cover ; {M}ultivariate normal ; density ; {S}pectral imaging data},
  booktitle = {},
  journal = {{I}nternational {J}ournal of {A}pplied {E}arth {O}bservation and {G}eoinformation},
  volume = {107},
  numero = {},
  pages = {102675 [13 ]},
  ISSN = {1569-8432},
  year = {2022},
  DOI = {10.1016/j.jag.2022.102675},
  URL = {https://www.documentation.ird.fr/hor/fdi:010084215},
}