David Da SILVA

• PhD Student in Computer Science
• Inria/Cirad/Inra project Virtual Plants
• Address:
  • TA A-96 / 02
  • Avenue Agropolis
  • 34398 Montpellier Cedex 5, France
• Tel: 33 (0)4 67 61 58 00 (ask for: 52 15) Fax: 33 (0)4 67 61 56 68
• email:David.Da_Silva_AT_sophia.inria.fr

• Context The plant's architecture is a basic concept used in botany since thirty years to describe the plant's structure organization. This concept refers to two principal types of information on the plant considered in space and time: the geometry (form) of its components on the one hand, and their topology (standard and adjacency of these same components) on the other hand. Many botanical works are dealing with the characterization of these two types of information for various species, varieties of plants and conditions of development. However, up to now there is no general method making it possible to quantify two fundamental and closely related topics of the plant's architecture study, whiches are :
  • The characterization of the irregularity (sometimes very strong) of the plant's shape(its leaves distribution in space for example).
  • The degree of similarity on each the scale of the ramified systems contained in a plant.

These two concepts have in common that they rest on an scale analysis of plants. Those techniques and concepts take their source in the developments of the fractal geometry. The aim of this thesis will be to develop methods of quantification for these two fractal's descriptor,whiches are fundamental for comprehension and modeling the plant's architecture.

• Plant's fractal geometry quantification Two complementary characteristics families from the fractal geometry will be studied. The first being the computation of fractal dimension making it possible to characterize the growth rate of the plant's details according to the scale (on wood and/or foliage in 3 dimensions). Varied estimators will be considered and compared on artificial data bases (for which one knows theoretical dimensions fractales) and real (obtained by digitalization 3D of real plants). The second family of methods will consist in developing algorithms for the lacunarity analysis of 3D plants. These methods will have to make it possible to characterize multi-scaled aggregation of the foliage and the size distribution of vacuums zones.

• Illustration The models of lights in plant's foliage do the assumption of a homogeneous leaves distribution in space. However, many species present an multi-scaled organization and irregular foliage for which this assumption is not checked. To avoid this problem, empirical parameters are usually added to the initial model to compensate for the errors. We will try to elaborate a modeling approach that will give a geometrical meaning to the model's parameters using fractal geometry tools. Moreover, the whole software tools tht will be developed during this PhD thesis will be integrated into the AMAPmod module of the platform ALEA.

• 2005 - 2006 : Imperative programmation - UM 2 - Licence1 (30h ETD)
• 2005 - 2006 : Algorithm & Imperative programmation 2 - UM 2 - Licence2 (34h ETD)
• 2006 - 2007 : Graphical Interface - UM 2 - Licence 3 (12h ETD)
• 2006 - 2007 : Imperative programmation - UM 2 - Licence1 (30h ETD)
• 2006 - 2007 : Algorithm & Imperative programmation 2 - UM 2 - Licence2 (30h ETD)
• 2006 - 2007 : Logic - UM 2 - Licence 2 (31h ETD)
• 2007 - 2008 : Algorithm & Imperative programmation 2 - UM 2 - Licence2 (30h ETD)
  • Notes TP