tanbul 2004 
AUTOMATED VERIFICATION OF A TOPOGRAPHIC REFERENCE DATASET: 
SYSTEM DESIGN AND PRACTICAL RESULTS 
A. Busch * *, M. Gerke ", D. Grünreich *, Ch. Heipke", C.-E. Liedtke*, S. Müller* 
* BKG - Bundesamt für Kartographie und Geodásie, Richard-Strauss-Allee 11, 60598 Frankfurt, Germany — 
(andreas.busch, dietmar.gruenreich)@bkg.bund.de 
* IPI — Institut für Photogrammetrie und Geolnformation, Universitit Hannover, 
Nienburger Str. 1, 30167 Hannover, Germany — (gerke, heipke)@ipi.uni-hannover.de 
© TNT — Institut fiir Theoretische Nachrichtentechnik und Informationsverarbeitung, Universitit Hannover, 
Appelstr. 9A, 30167 Hannover, Germany — (liedtke, mueller)@tnt.uni-hannover.de 
Intercommission Working Group H/IV 
KEY WORDS: Automation, Change Detection, GIS, Quality, Updating 
ABSTRACT: 
Quality of an official topographic reference dataset is important since it is the groundwork for many applications. Verification is 
classified as being part of quality management in this paper. Its main topic is the automated verification of a topographic reference 
dataset by means of orthoimages. The objects from the dataset are compared to an up-to-date orthoimage in order to obtain 
information on their quality. The main objects of interest are roads and built-up areas. As it is assumed that most of the objects in the 
database are correct, the strategy is to use the ATKIS objects stored in the DLMBasis as a starting point. Automatic image operators 
being able to detect the objects of interest use as much prior knowledge as possible from the dataset, such as contextual and 
geometrical information. By this means inconsistencies between the ATKIS objects and the image features can be detected. 
Positional accuracy of roads is checked as well. To organise the verification of the data independently from its capture, a semi- 
automatic working environment has been installed. In the subsequent interactive step the human operator just has to focus on the 
objects not being found in the automatic run. In this paper we will introduce the whole system. we then focus on the automatic 
components and their integration in a semi-automatic workflow. Therefore, we can give a detailed report on the system in its daily 
application as well as an evaluation of the results. 
1. BACKGROUND We do not want to go into detail about these specifications but 
start with a subdivision of quality measures into two categories 
Verification as referred to within this paper is part of the that are important for practical applications owing to the 
concept of quality management for topographic reference arguments given below: 
datasets as realized at BKG (German Federal Agency for ^ 4  |jogical consistency. i.e. consistency with respect to the data 
Cartography and Geodesy). Therefore, we first present the model. 
background of quality and quality management with respect to e consistency as regards content, i.e. consistency of data and 
geoinformation. reality within the scope of the model. 
1.1. Quality of Geoinformation In this paper we refer to the first category as logical consistency 
d S 3 since it is characterized by the fact that it can be checked 
Every application based on spatially referenced data or geodata without any comparison of the data to the real world. We can 
pene contain knowledec about their quality or at deast an perform a complete check of this category using solely the data 
ka of he CoDsequence of possible errors and the risk set without additional information. Only routines and 
associated with these errors. The level of knowledge about functionality within the database or the GIS are needed. Once 
quality needed by the user. differs widely depending en implemented, the inspection of logical consistency is performed 
applications and Specifications given. Many Consumers vill rate automatically. Format specifications, topological constraints, 
the quality of spatially Referenced data by its fitness for use and uniqueness of identifiers, and domains of attribute values count 
the effort needed for handling the datasets. In contrast, among the criteria for logical consistency. For the second 
Sompanits create Rew products bry processing geodata pr by category a comparison of data and reality is required. Basically 
tinkins their data to La referenced, datiset are interested in the comparison can be performed by means of current sensor 
Knowing whether the reference dat fulfil the warranted data or field work. A complete comparison of data and reality 
Characteristics. requires a lot of effort and cost. but in return it furnishes all the 
; : ; ; update information for the data. Consistency as regards content 
Data quality should be described by a certain set of measures, is the focal point of this paper. 
which express comprehensive and useful criteria. These should 
enable the user to compare the quality of different data sets. A well defined system is spanned by four quality measures, 
Therefore, quality measures are pan of standards. er namely completeness, correctness, consistency, and accuracy, 
specifications from e.g. ISO, CEN, or the OpenGIS Consortium. which are conceptually independent, i.e. orthogonal (Joos, 
  
Corresponding author. 
735