Recovering Data with Digital Photogrammetry and Image Analysis Using Open Source Software

• Main Author: Nielsen, Niels Ivar
• Format: Master Thesis
• Language: English
• Published: 2017
• Subjects: Open Source Software; Svalbard; Recovering Data; Image Analysis; Structure from Motion; Photogrammetry
• Online Access: http://hdl.handle.net/10852/58215; http://urn.nb.no/URN:NBN:no-60975

Old images are a valuable source of information in Geosciences. They provide information on the state of an object or phenomenon at a given time. Datasets of the past are produced according to old data standards optimized for analog management and analysis. Modern analysis and processing methods in digital systems are able to extract more information than what was possible earlier, but old formats and data standards make the transition to digital systems somewhat complicated. This thesis will present a workflow for creating orthophotos and digital elevation models from a set of high-oblique aerial images of Svalbard. Digital image analysis methods are applied to the images to locate fiducial marks and mask terrain features. Fiducial mark locations are used to calculate inner orientation for the images while terrain feature masks remove areas of the image that contain water, clouds or sky. \emph{Structure from Motion} is applied to the images to generate dense georeferenced point clouds. The last processing step generates orthophotos and digital elevation models that are useful for geoscientific analysis. Through the use of well known and transparent methods, with solid foundations in scientific literature, and by using only open source software, the work can be tested, reviewed and used by others.