Back to search

Democritus University of Thrace (DUTH)

Multi-camera systems for navigation of robotic systems and surveillance

Abstract

dc:description

During the last decade, vision systems established themselves as the standard in their own right and as one of the most efficient sensorial systems in industrial and everyday applications. Furthermore, it is apparent that one of the most challenging problems is the human-robot interaction inside the working area of a robot. Robots working in an indoor environment must be able to navigate seamlessly in the confined space and to avoid obstacles and humans in both static and dynamic environments. To achieve that, the vision system should enable a robust and effective 3D mapping of the working environment. Both of these fields, multi camera systems and environmental mapping, are the two pillars of this PhD dissertation. The motivation of this PhD dissertation has been the observation that the plethora of algorithms already developed by the scientific community in the field of computer vision for monocular and stereo systems are not able to provide sufficient feedback and adequate performance in case of a multi view application. Therefore, the objective of this doctoral research has been the introduction and implementation of such algorithms in multi camera networks for applications such as the environmental monitoring of the working space of a robot and its navigation using three-dimensional visual information. Innovative computer vision algorithms have been developed and evaluated for networks with minimum number of four and up to forty cameras. These algorithms were tested successfully on multi-camera systems targeting the maximum space coverage of static and dynamic environments and the three-dimensional reconstruction of everyday objects inside common workspaces. The thesis in hand is divided in three main chapters. The first one refers to the basic principles of multi-camera systems in terms of geometry and image correspondence. In the first section of the second chapter, the main focus is on the static, off-line camera topology problem, where the main goal is to manage the optimal positioning of a number of visual sensors for a region to be observed, given a set of task-specific constraints. A novel modified bio-inspired technique is proposed, which uses a bee colony algorithm that is able to determine the minimum number of cameras needed to cover the given space completely. The algorithm employs several camera placement constraints referring to geometrical, optical as well as reconstructive limitations. In the next section, an enhancement of the bio-inspired algorithm is presented to cover areas with arbitrary floor plans and significant dynamic crowd densities. The main contribution of this method is the development of a new optimisation technique for the maximisation of the coverage of a given space using cameras with ability to alternate their position resulting to a dynamic camera topology. In the third chapter, the first method encapsulates the classic problem of inferring a dense 3D volumetric reconstruction of an object from a collection of calibrated views along a common global coordinate system. The most noticeable novelty of this algorithm lies in the computational simulation of the behavior rules of ant colonies as an optimisation technique for further refinement of the final 3D models. Finally, the last section describes a novel contribution referring to the 3D reconstruction of prismatic objects based on the information provided by the vanishing points of the scene. This is the first time that such a 3D reconstruction approach is proposed. The efficiency of the proposed method relies on its low reprojection error and its ability to produce compact representations of the reconstructed objects.

Degree

thesis:*
Grantor dc:publisher
Democritus University of Thrace (DUTH)
Year dc:date
2013

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Chrysostomou, Dimitrios
  • Χρυσοστόμου, Δημήτριος

Subjects

dc:subject × 22

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/29280

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Chrysostomou, Dimitrios; Χρυσοστόμου, Δημήτριος. Multi-camera systems for navigation of robotic systems and surveillance. Democritus University of Thrace (DUTH), 2013. http://hdl.handle.net/10442/hedi/29280