Analyse et description de la morphologie foliaire : application à la classification et l'identification d'espèces de plantes

Abstract : Nowadays, automatic identification of plant species, by image analysis, has become crucial to maintain, standardize or deepen knowledge about the plant community. This thesis focus on the problem of automatic identification of plant species using leaf images. It addresses two main challenges: The first challenge is the large number and the high variability in foliar morphology across species. The second challenge is the intra-species variability which occurs locally at particular regions of leaves. To overcome these two problems, a return to botany and especially to leaf botanical concepts is established in order to define an automatic structuring of leaves at two levels: The first level concerns a categorisation scheme according to the botanical concepts “arrangement” and “lobation". The second level consists in decomposing leaves into semantic parts. The approach of the thesis is based on two key parts: In the first part, we focus on establishing this botanical-based structuring process by defining simple geometric properties correlated with botanical definitions and observations. In the second part, we investigate the feasibility and opportunities to integrate this structuring process in the identification scheme. Particularly, we make targeted researches in categories and we define specific part-based models.Experiments are conducted using the ImageCLEF 2011 and 2012 Scan images leaf databases. We compare our approach with respect to the reference identification scheme, applied on the whole databaseand using the entire images, and with respect to several methods referenced in the literature.
Complete list of metadatas

Cited literature [129 references]  Display  Hide  Download

https://pastel.archives-ouvertes.fr/tel-01804577
Contributor : Abes Star <>
Submitted on : Thursday, May 31, 2018 - 6:31:07 PM
Last modification on : Thursday, October 17, 2019 - 12:36:06 PM
Long-term archiving on : Saturday, September 1, 2018 - 3:06:03 PM

File

TheseMzoughi.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-01804577, version 1

Citation

Olfa Mzoughi. Analyse et description de la morphologie foliaire : application à la classification et l'identification d'espèces de plantes. Traitement des images [eess.IV]. Télécom ParisTech; Université de Tunis El-Manar. Faculté des Sciences de Tunis (Tunisie), 2016. Français. ⟨NNT : 2016ENST0028⟩. ⟨tel-01804577⟩

Share

Metrics

Record views

152

Files downloads

54