Habib, AymanEl-Sheimy, NaserAl-Gurrani, Hussein2016-07-052016-07-0520162016Al-Gurrani, H. (2016). Non-parametric Spatial-Domain Algorithm for Analysis and Mapping of Urban Scenes Using 3D Point Clouds (Doctoral thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/27122http://hdl.handle.net/11023/3099The world population is expected to grow to nine billion by 2050. With half of the current seven billion people living in urban areas, urbanization most certainly will continue in the future. Such a steep urbanization growth curve requires and justifies the growing interest not only in urban sustainability but also in sustainable development in general, which explains the growing interest as well in spatial information as it can provide sound support for decision-making. Geomatics engineering, which is the core of spatial information collection and processing, provides a multitude of tools for spatial information end-users, such as planners. The research of this dissertation developed a new geomatics-based tool for the urban data collection process, which is the first and most time-consuming step in any urban planning project. Using 3D point clouds obtained using state-of-the-art surveying equipment, which consisted of stationary lasers scanners, a mobile mapping platform, and a UAV, an automatic comprehensive classification algorithm was developed for urban scenes. This algorithm identifies and provides essential information about basic features that can be used in analyzing and understanding a surveyed urban scene, including 1) building locations, dimensions, and other design details; 2) road inventory (e.g., traffic signs, overhanging traffic signals, poles, and powerlines); 3) greenery details (e.g., individual trees, separated canopies, and bushes); and 4) identification of parked cars. The algorithm developed in this dissertation is comprised of four modules to process the input data. In Module 1 conducts ground detection and removal. Then, points clustering and characterization using principal components analysis is accomplished in Module 2. The data are further classified in Module 3 into four main categories: buildings, cars, road inventory, and greenery. Fine classification and analysis of the candidate classes and information extraction is conducted in Module 4, which consists of several subroutines for each process. The new algorithm was evaluated using four different datasets from different types of sensors and different urban sites. The experimental results revealed that the algorithm successfully classified the input datasets into the targeted classes and provided accurate information about the recognized objects.engUniversity of Calgary graduate students retain copyright ownership and moral rights for their thesis. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission.ArchitectureGeographyUrban and Regional PlanningRemote SensingArtificial IntelligenceEnergyEngineering--CivilEngineering--EnvironmentalRoboticsLiDARObject RecognitionUrban Designdoctoral thesishttp://dx.doi.org/10.11575/PRISM/27122