Skip to Main content Skip to Navigation

Caractérisation de la cocoteraie des Tuamotu à partir d'images satellites à très haute résolution spatiale.

Raimana Teina 1
1 PEQUAN - Performance et Qualité des Algorithmes Numériques
LIP6 - Laboratoire d'Informatique de Paris 6
Abstract : This Ph-D thesis concerns the characterization of coconuts plantations from the Tuamotu archipelago (French Polynésie) using multi-spectral IKONOS images. The first part of the thesis addresses the problem of coconuts enumeration. This enumeration is deduced from a individual segmentation of each coconut. We propose an original segmentation method which does not make use of the infrared channel. The image is first classified into two classes ``vegetal''/``non vegetal'' using a Bayesian classifier. After various preprocessing stages, a watershed segmentation is then applied using specific initial marker points. The algorithm has been validated by image interpretation and by a ground truth. The second part of the thesis is devoted to characterize the plantations in terms of ``artificial'', ``semi-artificial'' and ``wild'' plantations. Two approaches have been proposed. The first one uses a classification on image pixels. A study have been performed on a land cover problem in order to determine the best classifier of the literature and the relevant information to provide to classifiers. We show SVN combined with a textural description of the image gives the best result for the characterization of plantations. The pixel classification has a high computational cost due to the size of image to be processed. We propose an original and alternative method to classify the coconuts plantations analyzing the spatial configuration of each coconut coordinates. We build a statistical test to identify a wild plantation to a Poisson Point Process.
Document type :
Complete list of metadatas
Contributor : Dominique Béréziat <>
Submitted on : Monday, September 13, 2010 - 11:14:59 AM
Last modification on : Friday, January 8, 2021 - 5:40:03 PM
Long-term archiving on: : Tuesday, December 14, 2010 - 2:39:24 AM


  • HAL Id : tel-00516952, version 1


Raimana Teina. Caractérisation de la cocoteraie des Tuamotu à partir d'images satellites à très haute résolution spatiale.. Interface homme-machine [cs.HC]. Université Pierre et Marie Curie - Paris VI, 2009. Français. ⟨tel-00516952⟩



Record views


Files downloads