A Systems Approach for Point Cloud Data Management Open Access
Downloadable ContentDownload PDF
This dissertation assesses the current state of systems data management, particularly of point cloud data from light detection and ranging (LIDAR) reports, and proposes a full multi-dimensional data management system using custom software. The volume of collected point cloud data is growing exponentially as instruments for LIDAR data collection are becoming commercially available, primarily used by government agencies and commercial companies. LIDAR, which is an optical remote sensing laser scanner that measures properties of scattered light to find range and/or other information of distant targets, creates returns that are stored in point cloud form, a multi-dimensional dataset. The returns in the point cloud form each contain a location (X,Y, and Z) and select attributes (e.g., time, intensity, and frequency). Point cloud data for the same region can be collected through different methods, at different times, and for different missions, as well as be stored in multiple formats and locations. Up until now, data management of these point clouds in a particular area of interest has been a manually intensive task of discovering, retrieving and fusing the information together. There is no available single tool for an end-to-end, non-manual systems solution. This is partially because, unlike data from a terrain matrix or an image, point cloud data are not usually located at predefined, uniformly spaced grid locations. The volume and ungridded nature of point cloud data create a challenge in managing point cloud data from multiple collects efficiently and within a reasonable timeframe. This dissertation used an end-to-end systems approach to point cloud data management by integrating existing technologies for a robust solution. It built upon available technology, past research, subject matter expertise, and database and Geographic Information Systems (GIS) advances, to create an end-to-end (ingesting, cataloging, selecting/understand, fusion/combination, and identifying a dataset for an application) systems solution. The research illustrated that the most effective way to query databases is to gather information about the datasets: their geographical location, the date and time of collection, the method used to sense the data, the mission for which the data was collected, and the level of processing that has occurred on that data. This dissertation loosely linked web resources of point cloud data with custom software that was developed to examine and process identified point data to emulate a full integration of a point cloud data management system. This research also showed the feasibility of integrating point cloud capable database technology with GIS technology and fusion algorithms, and of using an optimized and robust version of specifically coded fusion algorithms, which were flightline extraction and change detection routines.