Range segmentation and registration for three-dimensional modeling of large scale urban scenes.

Title: Range segmentation and registration for three-dimensional modeling of large scale urban scenes.
Identifier: AAI3278424
identifier: 3278424
Creator: Chen, Cecilia Chao.
Contributor: Adviser: Ioannis Stamos
Date: 2007
Language: English
Publisher: City University of New York.
Subject: Computer Science
Abstract: The goal of our research is to build an automatic 3D modeling system for large scale urban scenes. In this thesis work, various novel algorithms have been developed for segmentation and registration, focusing on important geometric features existing in urban scenes: linear features, circular features, planar regions, and arbitrary objects. The algorithms include: edge-based segmentation, region-based segmentation, linear feature-based registration, circular feature-based registration, and segment merging. Given a large number of unregistered 3D range images of an urban site as an input, our system automatically, accurately, and efficiently produces a tightly merged 3D model. The 3D models produced by this system can be used in urban planning, historical preservation, and virtual reality applications.
Type: dissertation
Source: PQT Legacy CUNY.xlsx
degree: Ph.D.

Media: Range segmentation and registration for three-dimensional modeling of large scale urban scenes.