An approach for object‐based positional error model

Abstract

Rigorous and yet flexible models for positional errors of geometric features in GIS are crucial to the modeling of error propagation and the appropriate handling of these errors. Existing error models were derived and described based only on point features. This paper presents a new approach for extending those point‐based error models towards an object‐oriented one. For this purpose, the classical concept of the object‐based “Minkowski operations” is employed. Multiple line and polygon datasets of the same area were considered as observations from which the weighted‐average position of the features could be calculated. The resulting weighted‐average features can be used as the reference or “true” values in measuring the error of each observation dataset. Results of the experimentation have proven that the technique is very useful in extending and implementing existing conceptual error models into practical work.