Abstract :
[en] Geospatial data visualization, known as geovisualization, often provides great support in the comprehension of the geographic information. For instance, geovisualization is fully integrated in urban planning, from the exploratory phase (e.g., for facilitating the diagnosis of areas where something needs to be done) to the monitoring stage (e.g., for evaluating dynamics phenomena). Initially carried out in 2D, this process increasingly extents to the third dimension as a way to enhance the visual perception of the surrounding world, which is fundamentally 3D. However, this shift is not without pitfalls, both in the graphical design of 3D model and the viewpoint selection. To promote the use of the third dimension, it is therefore essential to propose new strategies that facilitate this transition. As a result, this thesis intends to contribute to 3D geovisualization field development via the first proposal that aims to formalize its process into a knowledge network including both an expression (i.e., visual stimuli) and content plan (i.e., semantic world). Following this formalization effort, we realized that moving to the 3rd dimension strengthens the role of the viewpoint, especially in order to enhance the visualization techniques relevance in achieving visual targeted purposes. Indeed, as the camera is no longer simply oriented in a classic top-down direction (which is usually the case in 2D), the point of view needs now to be configured; and, due to the 3D graphical representation, this configuration is complex, in particular because occlusion issues are inevitable. This is why this research mainly tackles the best 3D viewpoint selection issue via a geocomputational method that
vi
automates and optimizes its identification within 3D scene. Note that specific attention has been carried out to incorporate the solution into a global semantic driven visualization process. Eventually, the proposal is validated through the development of an experimental framework that aims to evaluate our solution for a given visual selectivity task: visual counting. For that purpose, an empirical test has been conducted with experts in the form of interviews using an online questionnaire.
