International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012 
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia 
    
AN INVESTIGATION OF AUTOMATIC CHANGE DETECTION FOR 
TOPOGRAPHIC MAP UPDATING 
Patricia Duncan! & Julian Smit? 
IThe Chief Directorate: National Geospatial Information, Department of Rural Development and Land Reform, Cape Town, 
South Africa 
pduncan@ruraldevelopment. gov.za 
2Geomatics Division, School of Architecture, Planning & Geomatics, University of Cape Town, Cape Town, South Africa 
Julian. Smit@uct.ac za 
KEY WORDS: Mapping, Change Detection, Classification, Pixel, Object 
ABSTRACT 
Changes to the landscape are constantly occurring and it is essential for geospatial and mapping organisations that these 
changes are regularly detected and captured, so that map databases can be updated to reflect the current status of the 
landscape. The Chief Directorate of National Geospatial Information (CD: NGI), South Africa’s national mapping agency, 
currently relies on manual methods of detecting changes and capturing these changes. These manual methods are time 
consuming and labour intensive, and rely on the skills and interpretation of the operator. It is therefore necessary to move 
towards more automated methods in the production process at CD: NGI. The aim of this research is to do an investigation 
into a methodology for automatic or semi-automatic change detection for the purpose of updating topographic databases. 
The method investigated for detecting changes is through image classification as well as spatial analysis and is focussed on 
urban landscapes. The major data input into this study is high resolution aerial imagery and existing topographic vector data. 
Initial results indicate the traditional pixel-based image classification approaches are unsatisfactory for large scale land-use 
mapping and that object-orientated approaches hold more promise. Even in the instance of object-oriented image 
classification generalization of techniques on a broad-scale has provided inconsistent results. A solution may lie with a 
hybrid approach of pixel and object-oriented techniques. 
INTRODUCTION 
The Chief Directorate of National Geo-spatial 
Information (CD: NGI), South Africa’s national mapping 
agency, is responsible for the official, definitive, national 
topographic mapping, aerial imagery and control survey 
network of South Africa. One of the responsibilities of 
the CD: NGI is the capturing and revision of 
topographical data into the national integrated database 
of geo-spatial information. The process of detecting 
changes to the landscape and updating CD: NGI's 
topographic database is currently performed manually, 
which is time consuming and relies on the knowledge 
and interpretation of the operator. 
The focus of this research is on updating topographic 
data for urban built-up areas, as these areas can change 
rapidly. An automated method of detecting changes to 
these areas is needed so that the topographic database 
can be updated regularly. The proposed method of 
detecting change is through image classification. In this 
paper we will compare various methods of image 
classification for the purpose of updating topographic 
databases. The change detection part of the research will 
come at a later stage once the most appropriate method 
of image classification has been decided on. It is 
envisaged that changes will be detected by comparing the 
newly classified data with the existing topographic 
vector data. 
The imagery used in this study is 0.5m resolution aerial 
imagery. Available image bands are red, green, blue and 
near-infrared. Existing vector data representing 
topographical features is the basis for measuring and 
comparing changes that are detected. 
UPDATING TOPOGRAPHIC DATABASES 
THROUGH IMAGE CLASSIFICATION 
PIXEL-BASED CLASSIFICATION 
Supervised classification 
Using the maximum likelihood classification method, 
Walter & Fritsch (1998) found that forests are recognised 
as homogenous and are well detected, while agricultural 
areas may show inconsistencies due to planting structure, 
but they could also be well detected. The water class was 
the most easily detected. Larger streets are recognised 
without significant problems, but sometimes there is 
confusion between pixels from the street class and pixels 
that represent house roofs due to their similar spectral 
characteristics. Pixels are only recognised as settlement 
areas if they represent house roofs, while other pixels in