Matching of image features and vector objects to automatically correct spatial misalignment between image and vector data sets
Degree GrantorUniversity of Canterbury
Degree NameMaster of Science
Direct georeferencing of aerial imagery has the potential to meet escalating demand for image data sets of increasingly higher temporal and spatial resolution. However, variability in terms of spatial accuracy within the resulting images may severely limit the use of this technology with regard to operations involving other data sets. Spatial misalignment between data sets can be corrected manually; however, an automated solution is preferable given the volume of data involved. This research has developed and tested an automated custom solution to the spatial misalignment between directly georeference aerial thermal imagery and vector data representing building outlines. The procedure uses geometric matches between image features and vector objects to relate pixel locations to geographic coordinates. The results suggest that the concept is valid and capable of significantly improving the spatial accuracy of directly georeferencing aerial imagery.