Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986 
Automatic digitizing of photo interpretation overlays 
with a digital photodiode camera: The ADIOS system 
C.A.de Bruijn & A.J.van Dalfsen 
ITC Department of Urban Survey and Human Settlement Analysis, Enschede, Netherlands 
ABSTRACT: The rapid introduction of geoinformation system in various fields of planning and municipal 
administration creates an urgent need for efficient methods to input information obtained by visual 
photointerpretation. 
The paper describes the ADIOS system for "automatic digitizing" of photointerpretation overlays, currently 
under development at the ITC Department of Urban Survey and Human Settlement Analysis. 
A digital camera with 2048 x 2048 pixel resolution is used to scan landuse interpretation overlays. After some 
processing to improve the quality of the scanned lines, the landuse codes of each polygon are added, from a 
centroid file or during an interactive session at a color graphics terminal. Resulting data can then be sent 
to a geoinformation system like USEMAP for geocorrection and further processing. 
1 INTRODUCTION 
Aerial Photography can supply planners of fast 
growing cities with thematic information on landuse, 
residential patterns, site suitability, traffic 
behaviour, etc.. Often such information is difficult 
to get by other means. Visual interpretation will, 
for the time being, remain the predominant technique 
for data extraction from airphotos, especially in 
urban areas with complex spatial structures. 
However, data obtained in this way have to fit in 
the geo-information systems that will increasingly 
be adopted by municipal authorities for their 
spatial datahandling, both thematic and topographic. 
Since 1973 various methods to input photo 
intepretation data in geodatabases have been 
developed at ITC as part of the USEMAP software. 
Starting with manual gridencoding they now rely 
mostly on manual digitizing. 
Valuable results have been obtained in various 
projects. The use is rapidly increasing and has 
created new possibilities in linking thematic survey 
data closer to their eventual applications. 
Still, manual digitizing is far from optimal and as 
pointed out by TOMLINSON (1980) digitizing one of 
the main problems in the field of geodataprocessing. 
Better, faster and more accurate input methods are 
urgently required if airphoto interpretation is 
going to remain an efficient datasource. 
2 VARIOUS METHODS OF DATA INPUT 
Entries in a geodatabase consist usually of two 
elements: a spatial description (location of point 
or area) and attribute information (thematic 
information about the point or the area, e.g. 
landuse). The spatial description of an area can be 
in the form of lines (vectors) defining the limits 
of the area, or in the form of gridcells (pixels) 
that are marked as belonging to the area. Choice 
between lines and gridcells has been the source of 
much debate in the past, but the development of 
modern hardware has made it rather irrelevant, since 
it has become fairly easy to convert from one 
datatype to another. However, at present "large 
scale" topographic databases are usually vector 
databases, while "medium and small scale" thematic 
databases tend to be gridcell databases. 
Interpretation data can be entered into a computer 
database by coding gridcells, (now rather obsolete), 
manual digitizing (the most common method) or 
scanning. 
Manual digitizing during interpretation may distract 
the interpreters’ attention and is certainly not 
feasible for all types of interpretation, while 
digitizing after interpretation means an additional 
step in the survey process, it may delay the final 
results and introduce additional errors. 
Either way, digitizing remains a time consuming 
obstacle, when data are to be used in a digital 
environment and at some places a tendency has been 
observed to use lower quality satellite Remote 
Sensing data instead of data from airphoto 
interpretation "because the data are already in 
digital format". The introduction of more efficient 
data input methods is therefore of paramount 
importance to ensure the continuing usefulness of 
airphoto interpretation. 
In 1981, based on earlier work by MEISNER (1981) the 
first author wrote a research proposal to develop at 
ITC a "digitizing machine" based on a high 
resolution CCD camera. The proposal called ADIOS 
(Automatic Digitizing of Interpretation Overlays 
System) was accepted in 1982 but acquisition of the 
required hardware was complex and time consuming due 
to the absence of representatives for this type of 
digital cameras in Europe. In October 1985 an 
Eikonix 78/99 digital camera arrived at ITC and 
development of application software and testing of 
various methods could eventually start in 1986. 
3 EMPHASIS ON PHOTO INTERPRETATION OVERLAYS 
The starting point of the ADIOS concept is that the 
normal visual stereoscopic photo interpretation 
should not be affected by subsequent digitizing 
procedures. The automatic digitizing should be based 
on the "interpretation overlay", the transparancy 
containing lines and codes which forms the end 
product of most regular interpretation jobs, (fig.l)