Automatic Georeferencing Of Non-geospatially Referenced Provisional Cadastral Maps

Abstract

This paper presents a methodology for the automatic georeferencing of provisional maps that have no geospatial reference grid. The methodology, which is a modification of the generalised Hausdorff distance registration measure, entails an iterative ranking process that consists of a set of qualitative statistical quantification to evaluate the correspondence of two datasets (pixel and object), overcoming data ambiguities, including outliers and noise. The automatic methodology presented can replace the preliminary requirement for initial manual or semi-automatic georeferencing of non-geospatially referenced data that are prone to error. It is a preliminary step in an overall data integration approach intended to enhance the geometric quality of such positional-approximated provisional maps to be used as qualitative spatial data infrastructure. A comparison of the proposed automatic georeferencing with the commonly used manual process showed better results, both in the number of point correspondences identified and in the final geometric alignment of provisional cadastral maps with a reference topographic dataset.