Urban areas change detection using DEMs generated automatically from high spatial resolution stereo satellite images

Resumen

In the spatial-temporal modeling for vertical change detection by a mathematical difference, Digital Elevation
Models (DEMs) are used to monitor changes in height values. The main aim of this paper is to improve the
accuracy of the change detection technique by applying a merge between DEMs of the same characteristics
and therefore achieve better quality results. The research was developed using data gathered between 2006
and 2011 in two test areas. Two robust DEMs were generated from these dates, which were subtracted to
detect altered areas. The DEMs have been automatically generated using the semi global matching method
(SGM) from high spatial resolution stereo images which were provided by the Ikonos and GeoEye-1 satellites.
All these images were previously subjected to a block adjustment in order to correct the geometric differences.
The proposed technique in comparison with the basic technique, has improved the correctness in 3D change
detection by reducing the number of pixels detected as false positives in 2,03% in Test Area 1 and 3.24% in
Test Area 2.