New methodologies and techniques on the development of virtual globes for mobile devices and the web

Abstract

La tesis aquí presentada estudia la generación de mapas 3D en dispositivos móviles y plataformas web. Se abordan varios temas relacionados con la implementación de sistemas de globos virtuales en estas plataformas. Debido a su naturaleza, la tesis abarca diferentes áreas, incluyendo arquitectura de software, computación gráfica, modelado y teselado de terreno y simbología cartográfica. La tesis comienza con la presentación proyecto Glob3 Mobile (G3M), plataforma de experimentación, implementación y transferencia tecnológica utilizada durante la conducción de la tesis. Posteriormente, se describe una arquitectura multiplataforma de gráficos, con diversas novedades destinadas a facilitar el uso de gráficos 3D en aplicaciones GIS. Se prosigue con el tema del renderizado y modelado de terreno multi-nivel-de-detalle, explicando la implementación seguida, basada en una variante de Geometrical Mipmapping, y las contribuciones en ambos campos. Diferentes implementaciones del LoD-test son estudiadas. Se hace un estudio es la navegación en el entorno 3D, estudiando los distintos esquemas de manipulación de la cámara virtual, basada en ángulos de navegación Tait-Bryan y en interfaces multitouch, y proponiendo un sistema de manipulación bi-puntero para una manipulación física del globo. Así mismo, se describen las implementaciones móviles de VR y AR usando dispositivos de mano. El renderizado volumétrico de la atmósfera basado en un modelo de Rayleight scattering ha sido también objeto de estudio a fin de obtener una implementación eficiente en móviles. Como última contribución, la tesis presenta un novedoso sistema de reordenación de marcas en pantalla basado en una simulación física 2D que evita efectos de abarrotamiento y oclusión de los mapas.

The thesis studies the field of computerized 3D mapping on mobile devices and web platforms. It addresses several topics related to the implementation of virtual globes systems on these platforms. Due to its nature, the thesis encompasses different areas, including software architecture, computer graphics, terrain modelling and tessellation and cartographic symbology. As the range of topics is diverse, each subject is enclosed on a different chapter. In order of appearance, these are the contents of the thesis: Chapter 2. The Glob3 Mobile project (G3M) is presented which is the experimentation and implementation platform utilized during the conduction of the thesis. The chapter describes software a novel modularized architecture and an original multiplatform development strategy. Chapter 3. A graphics multiplatform architecture is described. As it is portrayed, the software implements a graphics engine utilized in the G3M project, aimed to facilitate the use of 3D graphics to unskilled developers. The engine features a novel algorithm for the automatic selection of GPU shaders. Chapter 4. A study of the different alternatives to the multi-level-of-detail rendering and modelling of terrain is presented in this chapter. G3M follows a well-known hierarchical scheme similar to geometry mipmapping. Nonetheless, novel variations of the LoD test are presented. Likewise, the chapter proposes a “pluggable” software architecture of the rendering system, which allows a great degree of optimization and customization. Chapter 5. This chapter presents a particular version of the terrain rendering strategy, which precedes the architecture presented on Chapter 4. The core of this system is a novel PoV-based LoD test also introduced in the chapter. The efficiency of this simpler version of the rendering system is thoroughly tested on different platforms, both in terms of memory and computing consumption. The results show that the earth model is stable and meets the fps requirements even on low-powered devices. Chapter 6. Here, the problem is coarsening elevation models is addressed. A novel method based on the resolution of a linear system is proposed. The results of this approach are studied and compared with other alternatives. The results show that, for the assumed error metric, the method outperforms its competitors on several fields for the tested DEMs. Chapter 7. In this part of the document, the virtual camera system is described. Many definitions of the camera managing are given, including a parametric definition utilized in the implementation based on the Tait-Bryan navigation angles. Moreover, the chapter presents several schemes of interaction and user interfaces, with special emphasis on the multitouch interaction. As a novelty a methodology to achieve accurate interaction with the globe model by using the pinch gesture is proposed. Chapter 8. This chapter is a summary of the different aspects of presenting the virtual globe on VR and AR given the available mobile hardware. The achieved solutions to the implementation problems are explained in detail. Chapter 9. This chapter focuses on the atmospheric rendering effects, based on the Rayleigh Scattering equations. A study of this volumetric rendering is undertaken. A simplification of the model has been successfully implemented on GLSL. The study follows with the proposing of several novel ideas to achieve a more efficient implementation of the full scattering model. Chapter 10. A physics-based system for the on-screen reorganisation of cartographic markers for 3D virtual globes is proposed. The method successfully achieves its purposes, namely, the dynamic avoidance of cluttering and markers overlapping. Finally, the thesis summarized the different projects of technological transfer in which the novelties of this research have been successfully utilized.