A contour characterization for multiply connected figures

Abstract The characterization of a digital figure in terms of the multiple pixels, i.e., the pixels placed where the contour selfinteracts, can provide useful shape cues. These pixels can satisfactorily be detected if a suitable definition of contour simplicity is available. Such a definition is given in this paper, for the case of multiply connected figures. The multiple pixels are those where the contour fails to be simple, and they can be identified by using 3 × 3 local operations.