ON PERFORMANCE OF DIGITAL OFF-LINE SYSTEMS 
FOR AUTOMATIC PRODUCTION OF DTM DATA 
B. Makarovic 
Photogrammetry Department, ITC, The Netherlands 
Commission II, Working Group II-2 
ABSTRACT 
The main process stages are defined together with factors influen- 
cing performance, and sources of disturbances. The corresponding 
concepts and tools for theoretical assessment of performance and the 
quality indicators are also reviewed. The most crucial problem is 
assessment of parallax accuracy, which has not yet been solved sa- 
tisfactorily. The corresponding estimators can be verified by means 
of specific inputs with known matching characteristics. 
Fidelity of automatically produced DTM can be represented by a com- 
posed Transfer Funtion, and from it the critical frequency (terrain 
resolution) can be defined. 
I. INTRODUCTION 
The aim of the paper is to outline a framework for organisized 
assessment of the performance of systems for production of DTM data 
by digital off-line (time-delayed) technique. Such a framework and 
corresponding considerations make the problem area more transparent 
and thus, in turn, contribute to improved systems. 
The main process stages and the corresponding input-output data are 
defined in figure |. 
Scanning- Pre- Image Post- - 
| digitising | processing } matching A processing 4 
4 i | | | ! | 
! | y | | | | | | 
Input l Raw digital "' Upgraded Y Image Y Parallaxes, 
images images image data match DTM data 
Fig. 1 Process stages and data. 
Performance can be assessed by the following approaches: 
1. Experimantal tests using real data inputs or artificial data; 
2. Analytical, by means of theoretical tools; 
3. Semi-analytical, i.e., by a combination of experimental tests and 
theoretical analysis. 
This paper is concerned with theoretical analysis. The experimental 
approach was implemented by the ISPRS Working Group II/2, subgroup 
for Correlation Test. 
The contents of this paper are structured in two sections. The 
first, gives an overview of the influencing factors and sources of 
disturbances in the successive process stages. The second outlines 
the quality indicators for pictorial and geometric data, and it 
344