MENELAOS_NT – Multimodal Environmental Exploration Systems – Novel Technologies
Published July 14, 2021 | Version v2
Conference paper Open

EARTH OBSERVATION IMAGE SEMANTICS: LATENT DIRICHLET ALLOCATION BASED INFORMATION DISCOVERY

Authors/Creators

  • 1. Research Center for Spatial Information (CEOSpaceTech), University POLITEHNICA of Bucharest (UPB), Bucharest, Romania
  • 2. Earth Observation Center (EOC), German Aerospace Center (DLR), Wessling, Germany

Description

Land cover maps are among the most important products of Remote Sensing (RS) imagery. Despite remarkable advancements in land cover classification techniques, abundant detailed information in the very high-resolution RS images necessitates further improvements to harness the data and discover detailed semantic information. Moreover, scarcity of the labelled data and its quality is a major limitation in RS land cover mapping. In the present study, Latent Dirichlet Allocation is employed for semantic discovery in RS images and a novel kernel-based Bag of Visual Words model is proposed for land cover mapping.

Files

Earth_Observation_Image_Semantics_Latent_Dirichlet_Allocation_Based_Information_Discovery - MENELAOS Zenodo.pdf

Files (1.9 MB)

Additional details

Identifiers

DOI
10.1109/IGARSS47720.2021.9553122

Funding

European Commission
MENELAOS_NT - European Training Network (ETN) on Multimodal Environmental Exploration Systems – Novel Technologies 860370

Subjects

Data mining
https://id.loc.gov/authorities/subjects/sh97002073
Optical Detectors
http://id.loc.gov/authorities/subjects/sh85095144
Multispectral Imaging
http://id.loc.gov/authorities/subjects/sh85088388
Optical detectors
http://id.loc.gov/authorities/subjects/sh85095144