Multiscale project

MULTI-variate, -temporal, -resolution and -SourCe remote sensing image Analysis and LEarning

Funded by ANR (France) and Tubitak agency (Turkey)

MULTISCALE is a research project that aims at providing a complete and integrated framework for multiscale image analysis and learning with hierarchical representations of complex remote sensing images. While hierarchical representations of RS images has led to an effective and efficient scheme to deal with panchromatic or at most multiband data, their application to complex data is still to be explored. In addition, despite their ability to encode structural and multiscale information, their so far exploitation have not reached beyond a mere superposition of monoscale analysis. In this context, the MULTISCALE project defines new methods for the construction of hierarchical image representations from multivariate, multi-source, multi-resolution and multi-temporal data, and provides some dedicated image analysis and machine learning tools to perform multiscale analysis. The new methodology will be implemented in various toolboxes used by the community to favor the dissemination of the results. Success of the project will be assessed by benchmarking the proposed framework on two remote sensing applications. Substantial breakthroughs over classical methods are expected, both in terms of efficiency and effectiveness.

Objectives of the project

The MULTISCALE methodological objective is to propose new methods for the construction of one or more hierarchical image representations from multivariate, multi-source, multi-resolution and multitemporal data. These representations can then been exploited in order to perform feature extraction or to directly conduct multiscale learning of a region of interest. We intend additionally to explore the integration of a priori knowledge into the tree construction and description stages, when available (e.g. from ground truth or ontologies), with the end of better directing the underlying processes and thus improving the final results. In short, this project aims to provide a methodological foundation capable of adapting automatically to a given theme and region of interest requiring a multiscale analysis. In more detail, the project methodological objectives can be summarized as follows:

  • Extension of tree representations to multi-source, multivariate, multi-resolution and multitemporal data
  • Use of Full waveform LIDAR and optical data for motion estimation and classification
  • Integration of a priori knowledge into the tree construction phase
  • Multiscale analysis in an urban context of SAR and optical data
  • Computation of new features from trees
  • Disseminating the results through the integration in standard toolboxes
  • Multiscale classification

Publications

To be completed

Consortium

  • Equipe Obelix, IRISA, univ. Bretagne Sud
  • LETG, univ. Rennes 2
  • Gebze Technical University
  • Istanbul Technical University

Contacts

Please send all comments and questions on this project to Laetitia Chapel and Erchan Aptoula