GENLIB - Developing a generic framework for library based multi-site mapping of urban areas

You are here

Context and objectives

Extracting high-quality reference spectra for model training from remotely sensed imagery is a labour intensive and time-consuming process, especially in urban areas which are characterised by a high diversity of materials, as well as a high within-class spectral variability. Difficulties involved in training data selection, and the constantly increasing volume of remotely sensed data that becomes available urges the need for developing urban mapping approaches that are less dependent on site-specific reference data. The effort of having to collect and label reference spectra for each new application might be reduced by sharing image endmember spectra in open spectral libraries. The main aim of the GENLIB project is to propose a conceptual framework for the development and use of a generic urban library (GUL). This entails the definition of a suitable data model and metadata scheme for archiving urban image spectra, as well as the definition of a set of tools for dynamic use of the library.

Project outcome

Expected scientific results

  • A conceptual framework for the development of a generic urban library
  • A prototype version of a GUL management tool for handling multi-source/multi-resolution endmember spectra
  • Proof-of-concept of the GUL’s use for multi-site/multi-source/multi-scale urban mapping

Expected products and services

  • A generic class hierarchy for labelling of urban surface types
  • A metadata model for documenting and exchanging image-based endmember spectra
  • Tools for management/dynamic use of a generic urban library
Project leader(s): VUB - Cartography and GIS
Belgian partner(s)
SOMERS Ben KULeuven - Department of Earth and Environmental Sciences (EES)
Location: Continent:
  • Europe

  • Brussels (main test site), multiple secondary urban test sites (mainly Europe)

Related presentations: