Abstract:
This paper presents two new evaluation methods on
edge detectors. The first is non-contextual and it
concerns the evaluation of performance of edge detectors
in terms of detection errors. The second method is contextual
and it concerns the evaluation of performance
of edge detectors used in image compression. In both
methods, we studied the influuence of the image
characteristics and edge detector properties on its
performance. Seven detectors have been evaluated and their
performance compared.