LSA for Validation of Digitized Rectangular and Straight-Line Features

Abstract

The quality of digitized spatial databases directly affects many mapping applications and decision-making processes in geographic information systems (GIS). Because all information in GIS is, to some extent, related to spatial data, spatial accuracy plays a very important role. A large number of features in digitized databases contain rectangular and straight-line segment components. This is especially true for urban GIS databases, where rectangles and straight lines are the most common basic features. Automatic validation of such digitized components and features is part of the approach toward applying a quality control process before the databases are used in applications. This paper introduces a least squares adjustment (LSA) method for validating digitized rectangular and straight-line features. The method can be seamlessly implemented in digitizing software so that the digitized features can be automatically checked and corrected, and the spatial quality is enhanced. In addition, the overall digitizing quality indicator estimated from the validation procedure can be used to represent the spatial quality of the digitized features, although it is only an internal or relative quality measure. Such a quality measure is also critical to the uncertainty modeling in GIS. The results of an experiment using this validation method are given.