lative 
rrela- 
81-288 
tation 
photo- 
„22-33 
19 von 
ildern 
rsität 
n Lan- 
ittwer 
neine, 
1sses. 
3kunde 
trie. 
). 402 
Rela- 
75 
>thode 
irlag, 
etry. 
Koga- 
  
  
PROGRESS IN GIS CHANGE DETECTION ABILITY 
by 
G. Konecny and W. Schuhr 
University of Hannover 
Institute for Photogrammetry and Engineering Surveys 
Federal Republic of Germany 
Commission III 
Abstract 
Basic geometric algorithms, like heuristic polynomial equations, which are still in use, namely for the geometric rectification 
and/or transformation of remote sensing data into GIS systems, lack reliability. Due to the characteristic error propagation, heu- 
ristic polynomial equations may cause extreme distortions in GIS data, which, beside others, can become the reason for a com- 
plete misinterpretation of environmental phenomenas, in particular in view of Change Detection. On the other hand mathemati- 
cally strict algorithms, which are complex in handling and time consuming, often are not really required for practical purposes. 
Therefore it is the intention of this paper, to present reliable algorithms, suited for the geometric transformation of radar remote 
sensing data into GIS systems, which compromise between heuristic polynomial equations and mathematically strict approaches. 
Finally methods and results carried out for digital image rectification are dealt with, as well as qualitative aspects of the final 
products (radar mosaics and radar image maps ). 
Keywords: Algorithm, Change Detection, GIS / LIS, Radarblockadjustment, Radar Mosaic 
1. INTRODUCTION data is an interpolation problem. There are in 
principal 3 categories of algorithms to solve the geo- 
The main reason for geometric image processing is, metric problems for remote sensing imagery, 
to derive geometric correct positions of the pixels, including radar-images: 
which contain surface related information as - non-parametric interpolation methods 
greyvalues, in order to achieve a reliable geocoding t (physical) parametric methods and 
of geometric distorted remote sensing image data for - combined approaches. 
the correlation with GIS- or map-data. The interpolation process can be basedon —— 
Thus the full success of a radar mapping campaign - ground control point- and/or texture information, 
depends on the ability of a proper rectification of - housekeeping data (like airborne GPS etc.) and on 
geometric distortions, in particular to avoid - combinations of both types of data. 
misinterpretations due to mismatching within a Geo- Ground control point data is derived from known 
Informationsystem (GIS). numeric coordinate values, from GPS, from or maps 
Main objectives in this context are the providings of and from imagecoordinates of corresponding points, 
suited algorithms and software, necessary to achieve which can be verified manually, but increasingly by 
(digital) geometric precise radarmaps or radar or- interactive digital (relative and absolute) correlation 
thophotos, in view of techniques. 
- GIS Integration 
- mosaicing, 
- Sensor comparison, 2.1 Non-parametric interpolation methods 
- updating etc.. 
This task has been solved by geometric improvement This are namely 
of slant- as well as ground range radar image data, - polynomial equations, 
based on suited algorithms, which at least allow to - spline functions, 
calculate 3 dimensional and not only 2 dimensional - interpolation in a stochastic field, like moving av- 
ground control point coordinates. erage, weighted mean, linear prediction etc.. 
For SAR images of flat terrain 2 dimensional 
heuristic polynomial equations like 
2. RADARGRAMMETRIC APPROACHES 2 2 
AND RESULTS x a. es dd Ta NY $a t ta y 
2 2 
: . f d '=b +b x +b +b xy +b +b 
The transformation of imagedata into map or GIS- an y 0 1 2* 3 y a sl 
707