A directional and parametrized transition detection algorithm based on morphological residues

The boundary identification represents an interesting and difficult problem in image processing, mainly if two flat zones are separated by a gradual transition. The most common operators work very well for sharp edges, but fail for gradual transitions. In this work, we have done a characterization for gradient by a study of the opening residues. This characterization is useful to identify and classify sharp and gradual transitions between two consecutive flat zones according to a size criterion.