5 DATA INTEGRATION FOR SPATIAL DATA REVI- 
SION: CURRENT AND FUTURE RESEARCH AT THE 
CENTER FOR MAPPING 
Operationally, the problem of spatial data revision can be divided 
in three steps: (1) identification of local changes, (2) collection of 
data reflecting the changes, and (3) processing, removing, and/or 
merging of new data with unchanged old data (consistently) to 
generate up-to-date terrain representation. 
Table No. 5 
i Product Derivation 
  
  
  
  
  
  
| <6>= (ISGISLISN} 
  
: Te es 
  
  
  
  
  
  
  
  
  
  
To identify local changes, we are exploring the use of new raster 
images (for example, digital orthophotos) and vector data 
representing the old terrain (such as DLG-3 files) to develop a 
means to automate the detection of local changes. The idea is to 
use the information (spatial position and attributes) of the vector 
data and the cartographic rules to learn about the nature of the 
image representation. For example, unchanged elements of the 
road network will provide information about how these roads are 
portrayed on the raster images. This, together with the rules 
  
about the road network (for example, a new road is connected to 
an existing road), is being used to develop a partial or highly 
automated solution for new road detection. 
Once changes are detected, representative data needs to be 
collected. We propose to study different data and collection 
options such as conventional photogrammetric data, digital 
orthophotos and orthophoto quads, digital elevation models, 
satellite multispectral data, mobile mapping systems, and 
classical GPS. Currently, we have been studying digital 
orthophotos and the mobile mapping system developed by the 
Center for Mapping (GPSVan™). Figure No. 1 is an example of 
integration of these two datasets. 
We have found that these data sources complement each other 
well. Figure No. 1 shows a new road which does not appear in 
the digital orthophoto-image used to review the corresponding 
DLG data. In such a case, the most cost-efficient collection 
approach may be the GPS Van"M, 
We also found that digital orthophoto data, generally, are not 
enough for revision of all the spatial data carried by digital 
spatial databases or topographic maps (the nine coverages of 
Table No. 1). We are currently investigating, besides the mobile 
mapping systems data, the use of DEM, color, and black and 
white photographic images as complementary data sources. This 
specific research just started and no finding can be reported yet. 
The last operational step is to process, remove, and/or merge, in 
a consistent fashion, the new and the unchanged old data to 
generate a new terrain representation. 
  
  
  
  
  
Figure 1. GPSVan™ (white), Digital Orthophoto, 
and DLG data (black) 
As a first step of processing, removing, and/or merging the 
datasets, data will be transformed into a common reference 
datum and a common coordinate system. Then the dataset of the 
current terrain representation will be compared against the 
collected datasets, and a search for coincidental data (and 
differences) will be conducted. 
682 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996 
  
Re 
Ur 
De 
Ra 
Ra 
cor 
Sy 
De 
Ra 
get 
Int 
Pr 
Ra 
Ev; 
Sta 
Re