Interpreting and Representing Spectral Images Without Ground Truth

Presenter: Steven Le Moan - Colourlab, The Norwegian University of Science and Technology, Norway.

The MULTIP team of the IETR is all happy to welcome soon Steven Le Moan, Associate Professor at the Department of Computer Science, Norwegian Colour and Visual Computing Laboratory (Colourlab), the Norwegian University of Science and Technology (NTNU), Norway.

Steven will present some of his recent research advances in spectral imagery when used to separate distinct materials, especially in the case of unsupervised classification (clustering). Please find below the title and abstract of the seminar, and a short bio of Steven.
Séminaire

Intervenant

Steven Le Moan - Colourlab, The Norwegian University of Science and Technology, Norway

Résumé

Spectral imagery (hyper/multi) is used in many applications such as agriculture, ecology, or cultural heritage, to capture properties of objects and surfaces more finely than with traditional colour imaging. However, spectral images are notoriously difficult to interpret for both humans and machines. Patterns that exist in the data typically hide in clusters with very diverse characteristics (size, shape, dimensionality, density), which makes them difficult to recognise, especially without ground truth. This seminar will cover recent work done at the Colourlab and beyond on interpreting and representing spectral images without ground truth for such applications as remote sensing and cross-media reproduction.

Biographie

Steven Le Moan received the Ph.D. degree in image processing from the University of Burgundy, France, in 2012. He then worked as a Postdoctoral Researcher at the Technical University of Darmstadt, Germany, and the Gjøvik University College, Norway. Until 2021, he was a Senior Lecturer in electronics, information and communication systems with Massey University, New Zealand. He is currently an Associate Professor of color imaging at the Norwegian University of Science and Technology, Gjøvik, Norway. His research interests include multi/hyperspectral image analysis, color science, and visual perception.

Lien Zoom

Lien : https://univ-rennes1-fr.zoom.us/j/81426282897
ID de réunion : 814 2628 2897
Code secret : 352229