Evaluating the quality of OSM roads and buildings in the Québec Province
|Advisor:||Roche, Stéphane; Mostafavi, Mir Abolfazl|
|Abstract:||In the recent years, a new method of spatial data collection based on the public participation has emerged which is called Volunteered Geographic Information (VGI). In VGI projects, citizens voluntarily generate spatial content and share it with the rest of the community. OpenStreetMap is one of the most successful VGI projects that aims to create an editable, free map of the world. Citizens can participate in this project by creating new features, editing the existing ones or adding attribute values to the features. Attribute values are stored in OSM database as “key=value” pairs. There are three ways of participating in OSM: digitizing aerial images, importing GPS coordinates and importing a data set. No expertise in geomatics or GIS is required for participating in the project. Furthermore, there is no pre-defined rule for tagging and OSM participants can add any key or value to the features. Consequently, there are real concerns regarding the quality of OSM data. This research, propose to assesse the quality of OSM database based on the quality elements that are mentioned in ISO standard (ISO 19157:2013). Due to the limited access to reference data sets, only completeness, positional accuracy, attribute accuracy and shape accuracy are evaluated in this research. In the first step, all the measure that are proposed for assessing the quality of spatial data are reviewed. Then, the quality of OSM roads and buildings databases are assessed using these measures. Finally, the statistical analysis is done to measure the correlation between quality measures and potential quality indicators such as population density, income level and the distance to the center of the city. These indicators can provide an insight into the quality of OSM data where no reference data is available.|
|Document Type:||Mémoire de maîtrise|
|Open Access Date:||16 November 2020|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.