Improvements over the geometric correction of CBERS-CCD images by using classified georeferenced samples

The CBERS satellite images are freely distributed, but a geometric correction method is required to allow their use. We propose improvements over the automatic correction method by selecting the best reference samples from which control-points are generated. Further, those tie points will be used to define the geometric model used on the correction process. To demonstrate the effectiveness and performance gain we applied the correction method to a set of selected CBERS images using the automatically selected samples.

Keywords:
optimization; samples; selection; image; register; remote; sensing