Abstract :
There are many methods researched for extracting ground control point (GCP), however, most of them are based on flat areas or optical images. Moreover, there exist many shortages, such as huge quantities of calculation or strict restrictions in applications and so on. Anyway, they don´t suit SAR images under undulate areas. In this paper, a method of automated extraction of GCP under undulate areas is presented, using the characteristic of layover points and ridge points. First, extract control points in reference image and real-time image respectively by comparing their grey to surrounding points; then, filtrate and match them according to their correlation coefficient. Since a skipping mode of window gliding is selected for selecting GCPs, so the time extracting them is saved greatly. In addition, the round window is used to instead of the traditional rectangle window in matching, which enhanced its adaptability to the rotation and the grey and contrast change of image. In the end, a simulation according to a real and simulated SAR image is done, whose results proved the efficiency and accuracy of the method.
