Vous êtes ici : Accueil » Kiosque » Annonce

Identification

Identifiant: 
Mot de passe : 

Mot de passe oublié ?
Détails d'identification oubliés ?

Annonce

5 février 2020

Processing and analysis of hyperspectral images acquired by UAVs


Catégorie : Doctorant


PhD Thesis Subject 2020-2022

Processing and analysis of hyperspectral images acquired by UAVs

 

IETR – UMR 6164 CNRS / SHINE Research Group - Lannion

KEYWORDS

Unsupervised Machine Learning, Hyperspectral Imaging, Online Classification

 

REQUESTED SKILLS

Master M2 with strong skills in Signal and Image Processing, Statistics and Applied Mathematics, Machine Learning; Python, Matlab, R, C/C++ skills; fluency in written/oral Scientific English.

WHERE

The thesis will take place within the IETR team - Lannion site, University of Rennes 1/ Enssat, for a period of three years (2020-2022): tsi2m.enssat.fr - www.ietr.fr/spip.php?article1610

FOR MORE information, PLEASE CONTACT CO-SUPERVISORS:

Benoit Vozel, benoit.vozel@univ-rennes1.fr, 02 96 46 90 71
Claude Cariou, claude.cariou@univ-rennes1.fr, 02 96 46 90 39

SUPERVISOR

Kacem Chehdi, kacem.chehdi@univ-rennes1.fr, 02 96 46 90 36

 

BACKGROUND AND DESCRIPTION

Hyperspectral imagery sensors mounted on UAVs [1, 2, 3] have become privileged means for observing territories and their evolution [4] (temporal monitoring of urban and rural green spaces, greenways, algal deposition and invasive plants, crop diseases, roof mapping, etc.). Until today, the online processing capacity of the large volume of data they deliver has remained relatively limited. Most of the processing is done offline, i.e. once the acquisition mission has been completed.

Nowadays, the online processing and analysis of these data has become a major challenge for an efficient and appropriate dissemination to users of the enriched products resulting from their analysis [1]. In this field, the state-of-the-art still lacks methods adapted to this type of processing, particularly for hyperspectral images with large spatial and spectral dimensions [4].

As a continuation of the team's current work, we will focus on the problem of unsupervised and on-line learning of statistical distributions of hyperspectral data acquired by UAVs, which is a prerequisite step for the optimization of image processing (filtering) and analysis (partitioning of pixels into classes). In this approach, we do not wish to introduce a priori information in the methods to be developed, neither on the distribution models, nor on the number of classes to extract from the data (degradations, thematic content).

This research project will therefore concentrate on a methodological development specific to the problem of unsupervised online processing of hyperspectral images. The main goal is to allow their analysis in limited time and to offer decision support and response time, both of them adapted to the needs of the end-user.

The research work will initially be evaluated and validated on the basis of real aerial images acquired by the laboratory's current hyperspectral imaging platform within the framework of various partnerships (EDF, IFREMER and INRA AgroParisTech). The evaluation will then be extended to hyperspectral data acquired by UAVs on new sites. The developments carried out will be promoted through academic and/or industrial partnerships, in particular within the framework of the Centre Technologique Drone Ouest and the Anticipa Lannion technopole.

REfErences

[1] P. Horstrand et al., “A UAV Platform Based on a Hyperspectral Sensor for Image Capturing and On-Board Processing,” IEEE Access, vol. 7, pp. 66919-66938, 2019.

[2] Y. Zhong et al., “MINI-UAV borne hyperspectral remote sensing: A review,” Proc. 2017 IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Fort Worth, TX, 2017, pp. 5908-5911.

[3] T. Xiang, G. Xia and L. Zhang, “Mini-Unmanned Aerial Vehicle-Based Remote Sensing: Techniques, applications, and prospects,” IEEE Geoscience and Remote Sensing Magazine, vol. 7, n°3, pp. 29-63, Sept. 2019.

[4] Garrison J. (Edt.), “Hyperspectral imaging – From algorithms to physical models and applications,” IEEE Geoscience and Remote Sensing Magazine, vol. 7, n°2, 2019.

 

Dans cette rubrique

(c) GdR 720 ISIS - CNRS - 2011-2020.