Change Detection in High Resolution Satellite Imagery using Graph Theoretical Features

Detecting changes over a specified region is of major concern for government agencies, military, and rescue teams during a natural disaster. Government agencies can monitor buildings on government owned land automatically. Forests may be monitored for undesired changes this way. As for military, strategic plans may be updated based on the observed changes on targets. After a natural disaster, even if that region can not be reached, changes in satellite images provides necessary information to measure the effect of the damage. To solve all these problems, there are high resolution commercially available satellite images available. The important point here is to detect changes on these images automatically. In this study, we test graph theoretical measures for automated change detection. The performance of these measures are provided based on test images.