Improving the reliability of DTMs generated from high resolution satellite imagery by enhancing the automatic stereoscopic matching

Abstract EN

This paper presents a simplified method of epipolar resampling and a matching algorithm for automatic generation of Digital Terrain Model (DTM) from high-resolution satellite imagery.

Instead of the very complicated or very simplistic models for the epipolariry, the "quasi-epipolar lines" are calculated as simple circular arcs which lead to great simplification in the resampling process, reduction of time consuming and to an increase in reliability.

The matching algorithm severely reduces the problems of matching and provides dense, accurate and reliable results.

It leads to an impressive improvement of all accuracy indicators.

Using, the method proposed in this paper, we could achieve high accuracy and minimize time execution in the process of the DTM'e xtraction.

The results of an extensive test using a stereo SPOT model over South France are reported.

A strict sensor mathematical modeling is used here for accurate generation of epipolar images.

DTMs derived from the automatic matching with different options and the qualitative comparison of the results based on a sufficient number of check points and a reference DTM is presented.